A distinctive feature of contemporary marketing research, and more broadly speaking, of economic and social science research, is its advanced mathematization – understood here as the application of mathematical methods to capture the essence of some phenomenon. At its core, this process involves the permissible mathematical transformations that can be applied to a dataset, which determine the applicability of various statistical and econometric techniques with respect to the type of measurement scale. To facilitate the use of mathematics in drawing empirical conclusions from psychological data, which are often ordinal in nature, S.S. Stevens redefined measurement as “the assignment of numbers to objects and events in accordance with a rule” (Stevens, 1946). He introduced four fundamental types of scales that comprise measurement instruments – nominal, ordinal, interval, and ratio scales – and established criteria for the permissible statistical tests, methods and techniques that should be applied to each of them.

Stevens’ scales of measurement are still widely used in data analysis in the natural and social sciences, including marketing research. They were revolutionary, but they have certain flaws which have fueled an ongoing debate about the acceptability of using different tests and statistical techniques at different scales and levels of measurement (i.e. weak vs. strong scales). Instead of relying on Stevens’ scales, researchers may need to demonstrate the mathematical properties of their data and map them to analogous sets of numbers, making explicit claims about mathematization, defending them with proofs, and applying only those operations that are defined for that set (Thomas, 2019). Increasing mathematization can be explained by the needs of maximization, optimization, modeling, and forecasting.

However, we must ask whether the application of various statistical methods and techniques in marketing research has go too far, limiting the researchers’ horizon of thought, leading erroneous conclusions to be drawn, and diverting attention from trying to explain the non-quantitative attitudes, motives, opinions, needs, expectations, preferences of consumers (who are people, not machines or AIs). Addressing this question is the main goal of the article.

People’s attitudes are comprised of three closely related components: cognitive, affective and behavioral dispositions. Together these elements form a set of beliefs about the nature of the attitude object, making it difficult to establish clear boundaries between them at the measurement stage. Therefore, it is impossible to indicate where purely descriptive knowledge about the attitude object, or ideas about its nature, end and emotions and where assessments begin, where knowledge and assessments of the object end and where readiness, intention or sense of obligation to undertake specific behaviors towards the object (such as a purchase decision) begins (Nowak 1973; Escher 2010). This type of explanation is the key justification for the common practice of determining the direction and strength of attitudes only based on measurement of opinions expressed with varying degrees of acceptance.

The Likert scale is one of the most frequently used scales for measuring the direction and strength of attitudes of customers, consumers, and people in general. It was constructed to be applicable to measuring hidden phenomena (Likert, 1932) and was intended to overcome the limitations of simple scales, having the advantage of being multi-item. The method of appropriately using and analyzing data obtained on Likert-type measurement scales has been the subject of discussion for over 70 years. There are basically two main, competing views, which have evolved independently of each other, in the related literature and in the practice of empirical research. Historically, there has been a debate between those who support viewing the Likert scale in terms of ordinality (rank order – the present author is a supporter of this approach) and those who support intervalism – ascribing an interval-scale nature to the Likert scale (Burke, 1953; Glass, 1972; Walesiak, 1996; Kampen & Swyngedouw, 2000; Francuz & Mackiewicz, 2007; Jamieson, 2004; Carifio & Perla, 2008; Kaczmarek & Tarka, 2013; Kero & Lee, 2016). Supporters of the first approach point to Siegel’s argument (“the properties of an ordinal scale are not isomorphic to the number system known as arithmetic”, Siegel, 1956, p. 26), while opponents point to the authority of K. Pearson (1909), who pointed out that measurements on an ordinal scale can be treated as a certain version of measurements on an interval scale (for discussion, see Francuz & Mackiewicz, 2007, pp. 388–390).

The lack of a natural or arbitrary zero on the Likert scale creates a certain problem, so we do not know whether the distances on the scale are the same. For instance, is the distance on the Likert scale between “I completely agree”, coded as 5, and “I completely disagree”, coded as 1, equal to 4? Unfortunately, this remains unknown, because the “numbers” on this scale play a different role than they do, for instance, in the mathematical formula 5 – 1 = 4. The numbers on the Likert scale could be replaced, for example, with typographic symbols, such as emoticons. This fact determines the admissibility of using specific statistical methods and techniques in the process of data processing and inference. If a researcher calculates the arithmetic mean and standard deviation of data obtained on an ordinal scale, this is evidence either that they misunderstand measurement theory (whereby the type of scale is determined according to Stevens), or that they are implicitly assuming that the given scale is not an ordinal scale, but rather possesses interval properties.

Given that most of what is directly measured in marketing research – and often also in sociological, psychological, and even medical research – is measured suing ordinal Likert-type scales, a critical question remains: Is expressing ratings on an n-point “scale” (with 5, 7, or more points) truly a measurement on an ordinal scale? This issue extends beyond concerns about the “distances” between scale values; also pertains to whether a single numerical value can accurately represent a set of indistinguishable observations, as required by measurement theory.

2. The issue of interval or lack of interval of the Likert measurement scale

The Likert scale and its variants are situated on the ordinal level of measurement (Pett, 1997; Blaikie, 2003). This means that response categories, and therefore data obtained in ordinal-level measurement, are characterized by a rank order, and hence these empirical data can be compared and sorted. However, in research practice, these data are also often subjected to reduction processes, including latent variable analysis and correlation assessments, seeking to identify underlying factors that serve as the basis for empirical scaling and index construction. Yet it cannot be assumed that the intervals between values on the Likert scale are equal – although, as Blaikie (2003) points out, “researchers frequently assume that they are”. However, Cohen et al. (2000) claim that it is “unjustified” to conclude that the difference in intensity of feelings between “strongly disagree” and “disagree” is equivalent to the difference in intensity of feelings between other consecutive categories on the Likert scale. Nominal and ordinal variables (as well as interval and ratio variables) require different statistical approaches, and if an inappropriate statistical technique is applied, the risk of drawing erroneous conclusions from research findings (positive or negative verification of research hypotheses) significantly increases.

The scientific literature on statistics and research methodology consistently emphasizes that, for ordinal data, the median or mode should be used as a “measure of the central tendency”, rather than the mean. This is because the arithmetic manipulations required to calculate the mean (and the standard deviation) are inappropriate for data obtained by measurement on an ordinal scale, where numbers usually represent verbal statements (Clegg, 1998). Ordinal data can also be described using frequencies/percentages of responses in each category. Moreover, it is recommended that appropriate statistical inference for ordinal data be performed using nonparametric tests, such as Chi-square, Spearman’s Rho, or the Mann-Whitney U test, rather than parametric tests, because the latter require data at the level of interval or ratio scale measurement (Mann & Whitney, 1947; Lieberson, 1964; Myers, 2003; Sobczyk, 2007).

However, in practice these “rules” are often ignored by the authors of scientific articles, master’s theses, doctoral dissertations, and reports prepared by national and international research agencies. Such authors may, for instance, use a Likert scale, but describe and analyze the empirical data using means and standard deviations and conduct parametric analyses such as ANOVA. This is consistent with Blaikie’s observation that it has become common practice to assume that data obtained from a Likert scale measure can be processed like data obtained from an interval scale measure (at the interval level). In general, such authors do not clarify whether they are even aware that some would consider this to be invalid. There is often no explicit justification for assuming that Likert scale data has interval properties, nor any argument is provided to support this assumption.

3. Permissible operations on numbers depending on the type of measurement scales

In marketing research, in particular, the proper use of measurement scales is one of the basic problems. According to Stevens (1946), the permissible operations that can be performed on numerical data depend on the type of measurement scales used for the variables studied. Therefore, a different procedure is required when dealing with a data matrix that includes quantitative variables measured on scales of different types – i.e. when in addition to variables measured on strong measurement scales (i.e. interval and ratio scales), there are qualitative variables, characteristic of marketing research, measured on weak nominal and ordinal scales (e.g. data obtained from the measurement of attitudes, opinions, attitudes, preferences and expectations of recipients; product architecture and image; data from measurements of the color, quality and taste of products, packaging properties, opinions on the price level).

When all variables in a dataset are measured on a single type of scale, especially strong scales, the choice of statistical and econometric methods for analysis and interpretation is relatively straightforward. The problem of the transformation of measurement scales and permissible mathematical and statistical transformations for data obtained in individual types of measurement scales nevertheless often becomes apparent in social, economic and marketing research (Walesiak, 2014). What approach should researchers adopt, when specialist sources say one thing but common practice is different? The treatment of ordinal scales as interval scales, although common, has long been controversial (e.g. discussed by Walesiak, 1996) and – it seems – remains so. Kuzon et al. (1996) referred to the application of parametric tests to analyze ordinal data as the first of the “seven deadly sins of statistical analysis”. Knapp (1990), however, found some merit in the argument that sample size and distribution are more important than the level of measurement when determining whether it is appropriate to use parametric tests to assess specific parameter values for a given population from which the sample is drawn. These parameters may be the mean, variance or standard deviation.

Nevertheless, even if we accept that the status of intervals is justified in the case of data obtained using the Likert method, datasets generated using Likert-type scales often have a skewed or polarized distribution (e.g., when most respondents “agree” or “strongly agree” that a given brand of beer was tasty, or when respondents have polarized views on the “color of a beer bottle,” depending on their place of residence). Therefore, if we want to improve the quality of research in social sciences, and in marketing research in particular, such issues as the level of measurement and adequacy of mean, standard deviation, and parametric statistics should be taken into account already at the stage of research design, and authors must address them when discussing their chosen research methodology and the individual phases and stages of the research process, including specific activities, methods, and expected results at a given stage. Knapp (1990) proposed that researchers should decide what level of measurement is being used. To paraphrase: if data are measured on the interval level, for outcome x the researcher should be able to answer the question “x what?”. If the data are clearly ordinal, nonparametric tests should be used; and if the researcher is confident that the data can be reasonably classified as interval, attention should nevertheless be paid to the sample size, its representativeness, and whether the distribution is normal.

Finally, can we assume that Likert-type scales are interval scales? I remain convinced by the above arguments of Kuzon and Knapp. To paraphrase their reasoning: the average of “strongly agree” and “strongly disagree” is not “neutral and a half”, and this is true even when whole numbers are assigned to represent those who “disagree” and “agree”!

In the design phase and implementation phase of the research process, researchers must also resolve methodological issues. The basic distinction drawn is between qualitative and quantitative methods. The former are characterized by a holistic approach to the research object, treating it as an individualized entity and seeking to uncover the deepest possible research findings, understanding the very essence of phenomena being studied. Qualitative methods are therefore particularly suitable in social sciences, especially in marketing, for such purposes as the analysis of subjective customer experiences, the meanings of messages, the motivations and attitudes of participants in market exchange processes, or for holistically reconstructing or predicting the course of specific market processes (Bryman, 2005; Devine, 2006). Quantitative methods, on the other hand, are based on completely different logic, assumptions and research goals. When applying them, the researcher should accept that the obtained results will not be as deep as in the case of qualitative research, that certain nuances and subtleties will be naturally omitted, and that the studied phenomena will be treated aspectually and without an individualized approach. In exchange, the research results, i.e. the new information, may be more reliable and objective (not burdened with the subjectivity of the subject or object of cognition), unambiguous and precise in interpretation, while at the same time providing greater possibilities for generalization and, above all, making good decisions.

However, the potential benefits of the quantitative approach can be achieved only if the research is conducted carefully, methodically, and with strict control of the research process. The key element here is the measurement of variables. This is the thread connecting theoretical categories with empirical research and the means by which the former can be analyzed (Bryman, 2004). The essence of quantitative research is that the objects studied are not treated as holistic, ontologically separated entities, but as bundles of variables characterizing them. The main goal of quantitative research is to find relationships between these variables, through analyses revealing appropriate statistical relationships or their absence (Białas, 1999). However, in order for these analyses to be reliable and accurate, they must be based on input data of appropriate quality. Without this, they would be worthless, because statistics is only a tool and in itself cannot tell us anything valuable about market reality without solid work by the researcher and analyst.

This means that the element that determines the quality of the entire research process is measurement, understood as a sequence of research activities “aimed at determining the value of a specific quantity, and thus a numerical comparison of this value with a unit of measurement” (Szewczak, 2010). The activities that make up measurement may include the application of certain measuring tools, observation of their readings, as well as appropriate processing of directly obtained results – e.g. various calculations leading to determining the value sought. In short, “measurement is the assignment of numbers to objects in such a way that these numbers reflect the relations between these objects.” In the so-called representational approach to measurement, it is assumed that the measured properties are determined by means of empirical relations between objects, that can be characterized by them (Szewczak, 2010).

4. Measurement scales in measurement theory and properties of measurement scales according to Stevens’ classification

Measurement theory encompasses the entire scope of the measurement procedure, which also includes the construction of measurement scales, which serve as the instruments by means of which the value of a variable is measured. The researcher therefore performs an operation, by means of which the relations between certain objects can be observed, measured and interpreted. Regardless of whether we treat measurement scale construction as a separate research procedure or an integral part of measurement itself, it is one of the most important determinants of the reliability, validity and accuracy of quantitative study and to a large extent determines whether the results of study (useful information) can be considered valuable in the decision-making process. Only reliable instruments or measurement tools can ensure that the values of the variables subject to analysis correspond to actual characteristics of objects studied, and results of these analyses accurately reflect the structure of the market reality under study (Lissowski et al., 2008).

Constructing a measurement scale is not an easy task, it requires appropriate methodological competences and knowledge about the phenomenon or event being measured. It is also a time-consuming process. Therefore, in research practice, there may be a temptation to take shortcuts – omitting certain important elements, or even creating an ad hoc scale based on related indicators, selected according to the criterion of data availability, and then assuming that when summed up, these will jointly measure a phenomenon, event or process. Such an approach is not recommendable, because it leads to the creation of research artifacts and amounts to the mere simulation of scientific inquiry. A methodologically rigorous and reliable scale creation process, in contrast, ensures the reliability and credibility of the obtained instrument or tool. Although this process demands substantial effort and time, the benefits are significant: well-constructed measurement tools yield results that contribute to scientific knowledge and inform decision-making processes.

Researchers rely on multiple sources of information in their research, diagnostic, and prognostic endeavors, seeking to achieve both scientific and practical, utilitarian goals. A crucial part of this process involves selecting the appropriate measurement scales and research instruments to use. However, the measurement scales used must simultaneously meet several important criteria: (a) standardization, (b) reliability, (c) validity, (d) normalization, (e) feasibility of use (Stevens, 1956).

In marketing research, the concept of “scale” appears in three basic meanings (Sagan, 2003):

• in the relational sense, a scale defines the field of permissible transformations of sets of measured objects into a set of symbols while maintaining the principle of homomorphism, establishing a set of statistical analysis tools permissible for a given level;
• as an outcome of the research procedure, a scale defines the positions of respondents at discrete points or along the continuum of the measured feature (discrete-step or continuous variables);
• in data collection, a scale is a set of conventional categories or response patterns, in estimated graphic scales or so-called rank scales, which are instruments for collecting information and defining the direction and strength of respondents’ reactions to a given item within a complex measurement scale.

In the relational meaning of scale, the classification of measurement scales by the aforementioned S.S. Stevens (1946 and 1951) is adopted in marketing research methodology. This approach assumes that the type of measurement scale is known in relation to a given level of measurement. However, this distinction may be problematic for researchers in empirical identification, especially in relation to ordinal and interval scales. In contrast, researchers should have no problems distinguishing qualitative and quantitative data, discrete/step variables and continuous variables. The problems with the classification of Stevens’ measurement scales noticed in literature may be related to the fact that researchers may not recognize the type of scale a priori. The measurement operation is also related to theoretical construct adopted by researchers. The measurement procedure on Stevens’ scales, however, ensures access to data that are “empirically” at the appropriate level of measurement, and the transformation of variables that is mathematically and statistically permissible for a given level does not change their position at the points of the scale or its continuum (Townsend & Ashby, 1984; Mitchell, 1986). A measurement scale can also be treated as the result of a research procedure that determines position of respondents on a continuum (understood as a continuous, ordered set of an infinite number of elements that smoothly transition from one to another), or at points of measured feature. This is how the attitude scales of Likert (ordinal scale), Guttman (ordinal scale), and Thurstone (interval scale) are constructed and defined – the sum of the ratings for an individual respondent in relation to all items of a one-dimensional scale indicates the respondent’s position at points or on the continuum of the measured attitude, depending on its strength and direction.

In cases where a respondent’s position is determined by summing their individual scores across the scale, the result is essentially an attitude index (the scale is arbitrary in nature). However, when a respondent’s position is derived from specific mathematical procedures transforming raw scores (e.g. into factor values), then the resulting measure can be classified as an attitude scale (Sagan 2003).

Measurement scales are ordered from the weakest (nominal) to the strongest (quotient). In his foundational work, Stevens (1946) distinguished between intensive and extensive scales, emphasizing that the type of scale is associated with possible transformations that preserve its properties. The basic properties of Stevens’ measurement scales are presented in Table 1. The type of scale used to measure the value of a given variable (statistical feature), or more precisely, the properties of the chosen scale, determine the statistical methods that can be applied (Adams et al. 1965). The first two scales are classified as nonmetric (weak) scales, and the remaining two as metric (strong) scales.

It is important to recognize that the order of scales determines their level (power, strength). Nominal and ordinal scales are non-metric and qualitative scales, while interval and ratio scales are metric and quantitative. The metric scales are commonly treated in research together as a quantitative scale – this is the case in most statistical packages, including SPSS and Statistica. In experimental sciences, variables measured on a nominal and ordinal scale are most often referred to as discrete, and those measured on a quantitative scale as continuous. The distinction between measurement scales can therefore be summarized as follows (Wiktorowicz et al., 2020):

• When comparing the values of a variable expressed on a nominal scale (e.g. gender), we are only able to indicate whether two people have the same or a different variant of the variable.
• If we can additionally indicate which person has a higher variant of the variable (but we are not able to determine how much higher), we are dealing with a variable measured on an ordinal scale (this is the case, for example, with level of education or a feature measured on a Likert scale).
• If we can additionally indicate how much higher or lower a given variant is (distances are fixed), we are dealing with a quantitative scale.

And so, the stronger the measurement scale, the greater the accuracy of measurement, which in turn enables researchers to apply other advanced and complex methods of statistical analysis.

The data matrix is the starting point for mathematization and the application of statistical methods. The problem of applying, for example, multivariate statistical analysis methods becomes more complicated when variables in the dataset are measured on mixed scales or contain variables measured only on weak scales (especially on an ordinal scale). The problem of using methods like multidimensional statistical analysis, for example, occurs when variables in a data matrix are measured on non-metric scales.

The Likert scale is precisely such a non-metric scale, meaning it does not inherently possess the mathematical properties required for interval or ratio measurement. This raises the question of whether it is permissible to apply statistical tools designed for metric data to non-metric variables. One fundamental principle of measurement theory states that only measurement results on a stronger scale (interval, ratio) can be transformed into numbers belonging to a weaker scale (nominal, ordinal) (Steczkowski & Zeliaś, 1981;Wiśniewski, 1987; Walesiak, 1996, Jezior, 2013). Direct transformation of scales, consisting in their strengthening, is not possible, because from information Xn it is not possible to derive Xn+1 information or more (Walesiak, 1993). Whether mathematical manipulation of an empirical data matrix leads to valid research conclusions depends, among other things, on the validity of the initial mathematization of attitudes and the validity of the subsequent mathematization of empirical data, i.e. the permitted mathematical transformations, relations, and mathematical operations on these data. If attitudes are measured on an ordinal scale, respondents’ answers are only coded as real numbers, and mathematical operations are performed that are defined only for real numbers, not ordinal numbers, then these mathematical operations on the data matrix have no empirical equivalent and do not provide a basis for inferences or conclusions about attitudes.

If attitudes and perceptions exhibit the mathematical properties of real numbers and are limited, and statements offered on the Likert scale correctly define endpoints and consistent intervals on an attitude continuum, then there are two possibilities. The empirical data matrix can be mathematized as ordinal numbers because the data has mathematical properties of ordinal numbers, although this results in a loss of information.

However, the arithmetic mean and standard deviation cannot be calculated because they are undefined. Alternatively, the numerical values in the empirical data matrix can be real, and conversion of data into numbers involves rescaling. In this case, the numbers contained in data matrix are analogous to the object of study, operations are defined, and mathematical and statistical inferences lead to valid empirical conclusions.

5. Likert did not recommend calculating averages for data obtained on his scale

Rensis Likert, in 1932, cited Thurstone and Chave when he assumed that attitudes were formed on a linear “continuum of attitudes,” which was the basis for his explanation of how to construct a scale to measure attitudes (Likert, 1932). Likert proposed measuring attitudes based on respondents’ agreement with statements developed by researcher, the respondent marking various points on the “continuum” of attitudes. The statements should be arranged in order from one end of continuum to the other. Likert then explained that the statements should be assigned numbers, from one to five, in the case of a question with five options, with the number “one” being assigned to one end of the continuum and “five” to the other. Likert did not explicitly discuss the mathematical properties of these numbers, but he recommended calculating a correlation coefficient for each statement to ensure that the statement was numbered correctly, and he provided a table as an example. He treated the numbers of answers as if they were real numbers, and the continuum of attitudes as if they were limited (Likert, 1932, p. 50).

Likert did not recommend calculating average values, as is confirmed by this quote from his work:

The split-half reliability should be found by correlating the sum of the odd statements for each individual against the sum of the even statements. Since each statement is answered by each individual, calculations can be reduced by using the sum rather than the average. (Likert 1932, p. 48)

This, in turn, yields a clear answer to the question of whether the use of various statistical methods and techniques in marketing research has gone too far in empirical research on the nature of attitudes.

Returning to the discussion on the mathematical properties of the Likert measurement scale described earlier, this debate does not address the mathematical properties of attitudes themselves, on which the proper mathematization of the empirical data matrix depends. In fact, it is not even entirely clear whether attitudes can be ordered. There is ongoing debate in psychology, economics, and marketing about whether the evidence supports the idea that attitudes and preferences adhere to the principle of transitivity (if a > b and b > c, then a > c) (see, e.g., Regenwetter & Dana, 2011; Bleichrodt & Wakker, 2015), which is a property of both ordinal and real numbers. Additionally, Johnson (1936) raised early concerns about whether attitudes are dynamically stable. Whether various statistical operations are defined on Likert items and scales depends on how the empirical data matrix is mathematized. Performing operations that are not defined in mathematics is not mathematics – and as a result, it does not provide a valid basis for drawing empirical conclusions.

6. Conclusions

Given length constraints, this article concludes by proposing that future discussion should explore the following methodological issues regarding the incorrect treatment of different versions of the Likert scale as interval scales:

• the violation of the principle of equal intervals, which results from the principle of measurement isomorphism/homomorphism (especially “at the extremes” of the Likert scale, e.g. comparing distances 1–2 and 6–7);
• the validity of applying Thurstone’s method of successive interval scaling and other transformational procedures to Likert scales;
• the degree of suppression of Pearson correlation coefficients when calculated for Likert scales and the size of this suppression depending on the number of points – notably, 5–7 point scales are relatively resistant to the suppression effect;
• alternative measures and methods for analyzing multi-item Likert scales, such as using polychoric correlation coefficients instead of Pearson’s in the analysis of data with Likert scales (Sagan 2014).

The problem discussed herein is likely to become even more complex with the development of AI, machine learning, and data science and big data, because data scientists perform computational analysis but are not often involved in collecting the data or making decisions about how it is represented. They lack access to information about the empirical mathematical properties of the object of study, the evidence supporting the mathematization, and the set of numbers used, and moreover the programming languages they use may or may not allow for classification of the data by set of numbers or impose restrictions on the mathematical operations performed on the data with respect to type. This also encourages the treatment of all numbers as real, reducing the validity of empirical conclusions from the research process.

References

Adams, E. W., Fagot, R. F., & Robinson, R. E. (1965). A theory of appropriate statistics. Psychometrika, 30(1), 99–127.

Białas, S. (1999). Metrologia techniczna z podstawami tolerowania wielkości geometrycznych dla mechaników [Technical metrology with fundamentals of geometric dimensioning for mechanics]. Oficyna Wydawnicza Politechniki Warszawskiej.

Blaikie, N. (2003). Analyzing quantitative data. SAGE Publications.

Bleichrodt, H., & Wakker, P. (2015). Regret theory: A bold alternative to the alternatives. The Economic Journal, 125(583), 493–532.

Bligh, J. (2004). Ring the changes: Some resolutions for the new year and beyond. Medical Education, 38(1), 2–4.

Bryman, A. (2005). Research methods and organization studies. Routledge.

Bryman, A. (1988). Quantity and quality in social research (1st ed.). Routledge. https://doi.org/10.4324/9780203410028

Burke, C. J. (1953). Additive scales and statistics. Psychological Review, 60(1), 73–75.

Carifio, J., & Perla, R. (2007). Ten common misunderstandings, misconceptions, persistent myths, and urban legends about Likert scales and Likert response formats and their antidotes. Journal of Social Sciences, 3(2), 106–116.

Churchill, G. A. (2002). Badania marketingowe. Podstawy metodologiczne [Marketing research: Methodological foundations]. Wydawnictwo Naukowe PWN.

Clegg, F. (1998). Simple statistics. Cambridge University Press.

Cohen, L., Manion, L., & Morrison, K. (2000). Research methods in education (5th ed.). Routledge Falmer. https://doi.org/10.4324/9780203224342

Devine, F. (2006). Metody jakościowe [Qualitative methods]. In D. Marsh & G. Stoker (Eds.), Teorie i metody w naukach politycznych [Theories and methods in political sciences] (pp. 197–200).

Diamantopoulos, A., & Winklhofer, H. M. (2001). Index construction with formative indicators: An alternative to scale development. Journal of Marketing Research, 38(2), 269–277.

Escher, I. (2010). Pomiar kierunku i siły marketingowej postawy pracownika – kompromis pomiędzy teorią a praktyką marketingową [Measuring the direction and strength of employee marketing attitudes – a compromise between theory and marketing practice]. Acta Universitatis Nicolai Copernici, Ekonomia, 41(397), 159–174.

Fornell, C., & Bookstein, F. (1982). Two structural equation models: LISREL and PLS applied to consumer exit-voice theory. Journal of Marketing Research, 19(4), 440–452.

Francuz, P., & Mackiewicz, R. (2007). Liczby nie wiedzą skąd pochodzą. Przewodnik po metodologii i statystyce nie tylko dla psychologów [Numbers don’t know where they come from. A guide to methodology and statistics not only for psychologists]. Redakcja Wydawnictw Katolickiego Uniwersytetu Lubelskiego.

Frankfort-Nachmias, C., & Nachmias, D. (2001). Metody badawcze w naukach społecznych [Research methods in social sciences]. Zysk i S-ka.

Glass, G. V., Peckham, P. D., & Sanders, J. R. (1972). Consequences of failure to meet assumptions underlying the fixed effects analyses of variance and covariance. Review of Educational Research, 42(3), 237–288.

Główny Urząd Statystyczny. (n.d.). Pojęcia stosowane w statystyce publicznej [Concepts used in public statistics]. Główny Urząd Statystyczny. Retrieved March 1, 2025, from https://stat.gov.pl/ metainformacje/slownik-pojec/pojecia-stosowane-w-statystyce-publicznej/2924,pojecie.html

Hansen, J. P. (2003). CAN’T MISS – Conquer any number task by making important statistics simple. Part 1. Types of variables, mean, median, variance, and standard deviation. Journal of Healthcare Quality, 25(4), 19–24.

Jamieson, S. (2005). Likert scales: How to (ab)use them. Medical Education, 38(12), 1217–1218. https://doi.org/10.1111/j.1365-2929.2004.02012.x

Jezior, J. (2013). Metodologiczne problemy zastosowania skali Likerta w badaniach postaw wobec bezrobocia [Methodological problems of using the Likert scale in research on attitudes towards unemployment]. Przegląd Socjologiczny, 62(1), 117–138.

Johnson, H. M. (1936). Pseudo-mathematics in the mental and social sciences. American Journal of Psychology, 48(3), 342–351.

Kaczmarek, M., & Tarka, P. (2013). Metoda gromadzenia danych a ekwiwalencja wyników pomiaru systemu wartości w 5- i 7-stopniowych skalach ratingowych Likerta [Data collection method and equivalence of value system measurement results in 5- and 7-point Likert rating scales]. Handel Wewnętrzny, 5(346), 42–56.

Kaczmarczyk, S. (2014). Badania marketingowe. Podstawy metodyczne [Marketing research: Methodological foundations]. PWE.

Kampen, J., & Swyngedouw, M. (2000). The ordinal controversy revisited. Quality and Quantity, 34(1), 87–102.

Kero, P., & Lee, D. (2016). Likert is pronounced ‘LICK-urt’ not ‘LIE-kurt’ and the data are ordinal not interval. Journal of Applied Measurement, 17(4), 502–509.

Knapp, T. R. (1990). Treating ordinal scales as interval scales: An attempt to resolve the controversy. Nursing Research, 39(2), 121–123.

Krajewski, W. (1977). Correspondence principle and growth of science. Reidel.

Kuzon, W. M., Urbanchek, M. G., & McCabe, S. (1996). The seven deadly sins of statistical analysis. Annals of Plastic Surgery, 37(3), 265–272.

Lieberson, S. (1964). Limitations in the application of non-parametric coefficients of correlation. American Sociological Review, 29(5), 744–746.

Lissowski, G., Haman, J., & Jasiński, M. (2008). Podstawy statystyki dla socjologów [Fundamentals of statistics for sociologists]. Wydawnictwo Naukowe Scholar.

Likert, R. (1932). A technique for the measurement of attitudes. Archives of Psychology, 22(140), 55.

Mann, H. B., & Whitney, D. R. (1947). On a test of whether one of two random variables is stochastically larger than the other. Annals of Mathematical Statistics, 18, 50–60. https://doi.org/10.1214/aoms/ 1177730491

Mayntz, R., Holm, K., & Hubner, P. (1985). Wprowadzenie do metod socjologii empirycznej [Introduction to methods of empirical sociology]. PWN.

Mitchell, J. (1986). Measurement scales and statistics: A clash of paradigms. Psychological Bulletin, 100(3), 398–407.

Myers, J. L., & Well, A. D. (2003). Research design and statistical analysis (2nd ed.). Lawrence Erlbaum Associates.

Nowak, S. (1973). Teorie postaw [Theories of attitudes]. Państwowe Wydawnictwo Naukowe.

Nowak, S. (1985). Metodologia badań społecznych [Methodology of social research]. Państwowe Wydawnictwo Naukowe.

Pett, M. A. (1997). Nonparametric statistics for health care research. SAGE Publications.

Pearson, K. (1909). On a new method of determining the correlation between a measured character A and a character B. Biometrika, 7, 96–105.

Regenwetter, M., & Dana, J. (2011). Transitivity of preferences. Psychological Review, 118(1), 42–56.

Sagan, A. (2003). Skale i indeksy jako narzędzia pomiaru w badaniach marketingowych [Scales and indices as measurement tools in marketing research]. Zeszyty Naukowe / Akademia Ekonomiczna w Krakowie, 640, 21–36.

Sagan, A. (2014). Wprowadzenie do modelowania zjawisk społecznych i przykłady zastosowań w Statistica [Introduction to modeling social phenomena and examples of applications in Statistica]. StatSoft Polska.

Santina, M., & Perez, J. (2003). Health professionals’ sex and attitudes of health science students to health claims. Medical Education, 37(6), 509–513.

Siegel, S. (1956). Nonparametric statistics for the behavioral sciences. McGraw-Hill.

Sobczyk, M. (2007). Statystyka [Statistics]. Wydawnictwo Naukowe PWN.

Steczkowski, J., & Zeliaś, A. (1981). Statystyczne metody analizy cech jakościowych [Statistical methods of qualitative trait analysis]. PWE.

Stevens, S. S. (1946). On the theory of scales of measurement. Science, 103(2684), 677–680.

Stevens, S. S. (1951). Mathematics, measurement and psychophysics. In S. S. Stevens (Ed.), Handbook of experimental psychology (pp. 1–49). John Wiley & Sons.

Stevens, S. S. (1959). Measurement, psychophysics and utility. In C. W. Churchman & P. Ratoosh (Eds.), Measurement; definitions and theories (pp. 18–61). Wiley.

Szewczak, W. (2010). Jak zmierzyć demokrację? Teoretyczne i metodologiczne podstawy budowy skal demokracji politycznej w politologii porównawczej [How to measure democracy? Theoretical and methodological foundations for constructing democracy scales in comparative political science]. Przegląd Politologiczny, 4, 98–100.

Thomas, M. A. (2019). Mathematization, not measurement: A critique of Stevens’ scales of measurement. Journal of Methods and Measurement in the Social Sciences, 10(2), 76–94.

Thurstone, L. L., & Chave, E. J. (1930). Theory of attitude measurement. In L. L. Thurstone & E. J. Chave (Eds.), The measurement of attitude (pp. 1–21). University of Chicago Press.

Townsend, J. T., & Ashby, F. G. (1984). Measurement scales and statistics: The misconception misconceived. Psychological Review, 96(3), 394–401.

Walesiak, M. (1993). Statystyczna analiza wielowymiarowa w badaniach marketingowych [Multivariate statistical analysis in marketing research]. Prace Naukowe Akademii Ekonomicznej we Wrocławiu, 654.

Walesiak, M. (1996). Metody analizy danych marketingowych [Methods of marketing data analysis]. Wydawnictwo Naukowe PWN.

Walesiak, M. (2014). Wzmacnianie skali pomiaru dla danych porządkowych w statystycznej analizie wielowymiarowej [Strengthening measurement scales for ordinal data in multivariate statistical analysis]. Prace Naukowe Uniwersytetu Ekonomicznego We Wrocławiu, 327, 60–68.

Wiktorowicz, J., Grzelak, M. M., & Grzeszkiewicz-Radulska, K. (2020). Analiza statystyczna z IBM SPSS Statistics [Statistical analysis with IBM SPSS Statistics]. Wydawnictwo Uniwersytetu Łódzkiego. https://doi.org/10.18778/8220-387-5

Wiśniewski, J. W. (1987). Teoria pomiaru a teoria błędów w badaniach statystycznych [Measurement theory and error theory in statistical research]. Wiadomości Statystyczne, 11, 18–20.

Zeller, R. A., & Carmines, E. G. (1980). Measurement in the social sciences: The link between theory and data. American Political Science Review, 76(4), 996–1008.

New product competitiveness has important concern for industry and is decided by the interaction of marketing, finance, distribution and engineering activity with the market environment. This paper describes the characteristics of new product competitiveness in the market. The new product and market competitiveness in the field of new product strategy can be assessed at different levels-macro, meso, micro. This evaluation requires the use of appropriate measures, indicators and parameters (Juchniewicz & Grzybowska, 2010, p. 11). In particular, resolving the problem of measuring new product effects at micro level requires the availability of analytical tools that enable measurement (Wodecka-Hyjek, 2013; Hervas-Oliver, Sempere-Ripoll, Boronat-Moll, & Rojas-Alvarado, 2018). Information value and the effectiveness of monitoring of new product strategy and new product development process (NPDP) will depend on the adequacy of the metrics and parameters used. The most frequently and widely used are financial measures, on the basis of which companies exercise management control over organisational efficiency (Reinertsen & Smith, 2001; Carboni & Russu, 2018). Revenues, profits and other financial effects can be subject to manipulations (reducing companies’ expenditure on research and development [R&E] and marketing, and falsification and concealment of information). So the problem concerns what happens over time when the effects of these 'savings’ reveal a drop in new product competitiveness, and there is a decrease in the level of success of new products introduced on the market (Rutkowski, 2007).

Although expenditure on R&E is a key indicator of innovativeness, scientists have found ambiguous results regarding its effect on new product strategy and company performance. Researchers claim that variations in R&D effectiveness can be explained by changes in a company’s social system, in its new product and innovation management process. It is still unclear how innovation management influences R&D effectiveness in terms of NPDP and its maturity (Heij, Volberda, Van den Bosch, & Hollen, 2020). Managers, in response, took regulatory actions to increase new product market security and ensure sustainable development (Wheelwright & Clark, 1992; Cooper, 1993, 2019, 2021; Wu, Kefan, Gang, & Ping, 2010; Walker, 2013).

Reports from the literature on new product strategy effects, and the associated success and failure factors, contain conclusions describing critical and important issues in the NPDP. Empirical research reveals that the new product development (NPD) success rate is still at a low level and it depends on the level of NPDP maturity (Crawford, 1979; Griffin & Page, 1996; Stevens & Burley, 1997; Cooper & Edgett, 2008; PDMA, 2012; Lee & Markham, 2016; Rutkowski, 2022).

Efficiency in the area of new product strategy is defined by multiple factors, determinants and parameters; among these, one of the key ones is the competitiveness of the new product that is produced and distributed. Ultimately, in the aggregate of things, a competitive new product affects the competitive position of the company on the market.

Use of market share or relative market share measures are traditional approaches to defining new product competitiveness or market position. These measures to some extent characterise state of a company and its competitive practices. An important role in defining a company’s situation is played by the level of novelty and quality of the new competitive product. Distinguishing participants by the market share enables them to be assigned certain roles, e.g. leader or pretendent (Tyunyukova, Ruban, & Burovtsev, 2018). The results of empirical research also indicate that the market and innovation orientation is positively related to market performance of a new product. The results also show that NPD performance is highest when the market orientation and maturity of the relationship network are at a high level, which supports the proposed three-way interaction (Mu, Thomas, Peng, & Di Benedetto, 2017). So, the new product competitiveness levels for companies of different industries serve as the key factors for their success or failure on the market.

The Measures of New Product Strategy in Competitive Market

The difficulties in developing an industry standard for the success or failure of a new product can be cited. There is no appropriate consistency in defining the new product success or failure level. Each new product must have a specific strategic goal or goals. After fulfilling this goal, e.g. it can be removed from the product line. So, it was a strategic success from company’s point of view, not a failure, even though the new product was removed from the market, and it may be said that the removal would then be part of an overall strategic development.

Since the 1960s, various phenomena in NPDP have been analysed in the available literature, and the focus of research is shifting from defining the right process to ensuring its proper implementation, better management, better measurement and continuous improvement
(Rutkowski, 2007, 2021). The key factor of new product success is choice of the right new product strategy under the appropriate marketing strategy and presence of competitive advantage related either to the company or to its new product. Managers should take steps to ensure that such an advantage is effectively protected by copyright or patents, and that there are instituted measures that would effectively prevent, or at least deter and discourage, competitors from easily copying it. The competitive advantage of a company’s new product can be represented in many ways, but most of all it consists in the features or attributes of the new product, unnoticed by competitors (Thompson, Peteraf, Gamble, & Strickland, 2016). New product competitiveness is the parameter that indicates presence of such peculiarities and allows forecasting the future success of marketing activities of a company related to a new product strategy.

Table 1 presents selected goals of a new product strategy, which are of different financial, marketing (market) and technological nature and contain criteria in the sense that they determine the overall level of a new product’s success or failure on the market. On the other hand, the level of a new product’s success or failure on the market should be treated as a general measure of the company’s competitiveness level in the field of new product strategy, in particular the effectiveness of NPD strategy. Table 1 includes both the measures of competence and
competitiveness in achieving specific goals in NPDP (House-Price model), and the measures related to the level of achieving goals of new products after the period of commercialisation and introduction on market (Rutkowski, 2021).

Thus, currently provided research and available knowledge point out that there is no ideal process for developing a new product, as evidenced by the longstanding relatively high rate of new product failure or partial failure on the market (Castellion & Markham, 2013; Rutkowski, 2016, 2022; Cooper, 2017). Table 2 presents new products’ success or failure rates by sector/industry. For example, consumer goods and services show a lower rate of success than the software or healthcare sectors.

The empirical studies conducted so far indicate that the implementation, realisation and control of a new product strategy should be undertaken with particular care in a company, in order to ensure efficient management of the new product offer, including appropriate coordination, integration and communication link with the existing business processes, and to achieve the goals for which this strategy is implemented (Engelman, Fracasso, Schmidt, & Zen, 2017). However, a major problem in determining the success of individual outputs is the multidimensionality of NPD results (Rutkowski, 2016).

Quantitative Rates to Assess Success Level of Project Teams in New Product Strategy

New product competitiveness or product competitiveness has been the subject of debate by academics and practitioners over many years (Takei, 1985; Oral & Kettani, 2009; Roostika, Wahyuningsih, & Haryono, 2015). There are some aspects or problems that have not been resolved due to the differing purposes of research. The question of quantitative assessment of new product competitiveness is always relevant for companies in order to determine new product strategy and functional marketing strategy, in that it enables them to arrive at decisions concerning an increase or expansion in their market positions.

An analysis of scientific publications reveals some methods and concepts for the evaluation of new product competitiveness. The main approaches may be summarised as follows (Shpak, Seliuchenko, Kharchuk, Kosar, & Sroka, 2019):

(1) Methodology of new product competitiveness evaluation through calculating its ranks and weights. According to this approach, a new product rating is dependent on the quality indicators of the new product.
(2) Methodology of the evaluation of new product competitiveness through the volume of sales. This approach assumes that the volume of sales reflects market demand (customer demand), which is why it might be the most significant criteria for the product’s competitiveness. Under modern conditions, a high volume of sales might be the result of a weak competitive environment and the absence of similar new products on the market.
(3) Methodology of the evaluation of new product competitiveness through a complex index with multiple variables. According to this approach, a complex index of new product competitiveness should include a set of partial indicators that generalise the following characteristics of new product competitiveness: customer requirements, technical
requirements and enterprise expenses.
(4) Methodology of the prediction index of competitive strength of brands based on fuzzy logic, which is based on expert knowledge bases (quality of the brand product, image of the brand product, and service connected with the brand product).

When assessing a new product strategy effects, the most useful financial measure is profit level set for a given project. Among classic marketing rates, measures of level of satisfaction and acceptance of the new product offer by clients, as well as market share index, are characterised by relatively high usability. Nevertheless, assessment based on profit level offers a useful measure reflecting the efficiency of the development and commercialisation processes and efficiency of a new product on the market.

The measures of success or failure in implementation of a new product strategy, which are used by companies, are reactive in nature. Marketing decision support systems should be more proactive from managers’ point of view (Rutkowski, 2021).

Quantitative indicators can be the basis for analysis of economic, financial and technological potential and the innovative marketing strength of companies. The values of these indicators also reflect the directions of enterprises’ conduct in NPDP and technologies, as well as in other areas of innovative activity (processes, management, organisation, sales and marketing). Moreover, these indicators inform managers about the strength of linking marketing, finance and technology with the effectiveness of new product strategy.

In last decade, a new company paradigm has arisen, which assumes that resources of marketing, organisational and technological knowledge are in central importance for the value of a company. This new way of thinking makes it necessary to formulate a new product strategy at three levels (Kasprzak & Pelc, 2012):

• shaping the company’s competencies, reflecting technological knowledge resources;
• R&E, being the sources of knowledge and new technological solutions for products and processes; and
• mastering marketing, technological processes, product manufacturing and distribution systems as tools of competition and competitiveness.

These three levels of new product strategy formulation require prognostic recognition, strategic analysis of a different nature and efficient marketing decision support systems.

Measures of New Product Competitiveness as Effects of New Product Strategy

In the scientific publications, as well as in statistical sources (Commission Regulation [EC] No. 1450/2004 of 13 August 2004), among many indicators used to compare and determine the change trends in marketing and technological strategies of various industries and companies, the following measures of enterprises’ effects in the area of new product strategy and intellectual property protection are particularly useful (OECD/Eurostat, 2018; Grzegorczyk & Głowiński, 2020):

• innovative activity intensity (IAI), research and development intensity (RDI), new products marketing intensity (NPMI);
• engagement to innovation (IE), R&E and marketing of new products (NPME)-the degree of development of intellectual resources in relation to production investments;
• new product sales index-marketing product offer renewal (NPS);
• patent activity index (PAI); and
• inventive activity index (IVAI).

Strong changes in these indicators reflect problems with market position stabilisation of a company or the industry. The author assumes that the measure of intensity of competition on the new products’ market is the indicator expressed by the formula (Rutkowski, 2021):

where SR_(u,s) denotes the rate of competition (according to the share of sales [u] or sales [s]) in the new products market; SB_NPi the share of sales of the i-th new product (type, line) in the industry in the total sales value of products in a given industry (or WSB_NPi the sales value of the i-th new product [product type, product line] by all companies in a given industry); and SP_NPi represents the share of sales of the i-th new product (type, line) in a given company in the total sales value in a given industry (or WSP_NPi the sales value of the i-th new product [type, line] in a given company).

Additionally, SB_NPi – SP_NPi > 0; WSB_NPi – WSP_NPi > 0; and SR_(u,s) < 0.5, 0.5 ≤ ≤ SR_(u,s) < 0.65, 0.65 ≤ SR_(u,s) < 0.80, 0.80 ≤ SR_(u,s) < 0.95, and 0.95 ≤ ≤ SR_(u,s) < 1.0, respectively, indicate very low, weak, average, strong, and very strong intensity of competition on the new products market.

The index of the intensity of competition of a new product offer on the market (SR) reflects the level of competitiveness of a new product offer of a given company or industry. It basically determines the ability of the company’s new market offer to participate in smooth adjustment processes in the changing market conditions. It shows how companies compete on the market of new products for the favour of customers, and is helpful in indicating the degree of customer acceptance of the new product offer. Therefore, this index shows the ability to survive on the market, as well as the ability to develop the company under certain conditions of competition (Rutkowski, 2021). Information about the value of new products sales in the market is generally more accessible than information about the quantity of new products sales, which is very difficult to obtain.

To empower itself to undertake projects involving new products with a high probability of success in the future, the company must monitor internal NPDPs and the situation prevailing in the marketing environment. Thus, it is important to determine the metrics/indicators/parameters that can be used by these companies to measure the competitiveness level of a new product. The use of indicators tends to bring better analytical capacity in management information system and marketing decision support system, regardless of the target industry (Zizlavsky, 2016).

The parameter of new product competitiveness on market is a much broader concept, and it may vary significantly for the same new product quality, depending on the market state, competitors’ activities, their marketing strategies and appearance of new products in the studied product line (Tyunyukova et al., 2018). Nevertheless, price, costs, quality and technical level of the new product are inherent parts of competitiveness and shall be accounted for in its evaluation. However, the overall new product attractiveness is defined by the customer. The main groups of parameters accounted when evaluating new product competitiveness are:

1. Economic (costs associated with purchase, transportation to the place of operation, installation, and irrecoverable customs duties and other taxes that have to be borne);
2. Quality systems standards (compliance with the existing regulations, compliance with technical specifications, compliance with the contract, compliance with the standards — ISO);
3. Technical (product parameters and properties, usability, service life); 4. Operational (reliability, service ability, raw materials, other materials, electric power, maintenance);
5. Ergonomic (hygienic, anthropometric, psycho-physiological, psychological); 6. Aesthetic (harmony of shape, rationality of shape, preservation of saleable condition, quality of fabrication); and
7. Marketing (price level, efficiency of distribution, effectiveness of promotion and communication systems, brand strength and
awareness).

In the process of managing the competitiveness of new products on the market, an important issue is to establish the degree of influence of each of factors on the level of competitiveness. Therefore, the selection of the most important of them at a certain stage of the integrated product’s life cycle not only allows its quality to be improved but also enables changes in market conditions with respect to properties and characteristics to be considered, reduction in production costs and improvement in the price-quality ratio (Vashkiv, 2020). The results of a study of customer needs and market requirements constitute the basis for assessing the competitiveness of new products.

Based on the work of Tyunyukov and others, the author proposes the following formula for determining the competitiveness of a new product on the market. Actions are related to selection of the above-mentioned parameters of the evaluated new product, selection of the reference sample with ideal parameters and comparison of the surveyed sample with the reference one. All methods applied for comparing the samples can be categorised by the parameter of qualitative or quantitative research of competitiveness. Combined evaluation methods are represented by assessments (obtained from experts and customers by surveys) transformed into quantitative parameters using mathematical and statistical tools. Quantitative assessment is usually performed by calculating single, group and integral indices (Gabrusewicz, 2014). The subsequent actions in analytical process are related to selection of meaningful parameters of the new product in question, selection of the reference sample with ideal parameters and comparison of the researched sample with the reference one. All methods applied for comparing the samples can be categorised by the parameter of qualitative or quantitative evaluation of competitiveness. Combined evaluation methods are represented by assessments (obtained by means of surveying experts and customers) transformed into quantitative parameters using certain mathematical and statistical tools.

Quantitative assessment is usually performed by calculating single, group and integral indices.

New product single index is defined by the formula:

where NPSI represents the single parameter index; Pl the parameter level for the surveyed new product; and PL100 the parameter level for the reference new product sample, which satisfies the need in 100%.

After finding single parameter indices, a group parameter GPi can be calculated by Formula (2), which enables single indices to be integrated for a uniform group of parameters-economic, standard, technical, operational, ergonomic, aesthetic and marketing. Single indices can be integrated using factor weight set during customer or expert surveys.

where NPGP represents the group parameter; F_i the factor weight in group parameter (e.g. marketing : price); and g_i the single index.
Integral index is calculated in the format 'selected group of parameters, e.g. technical-tech/marketing-mark parameters’ (Formula [3]), which in fact provides the parameter evaluation in relation to its marketing characteristics.

where Gtech represents the group index for group of technical parameters; and Gmark the group index for group of marketing parameters.
Then, the conclusion is made: if NPH < 1, then the studied new product is inferior to the reference sample, and in case of NPH > 1, the reference sample has higher competitiveness. A significant drawback of such approach consists in the fact that only those new product parameters can be used for comparison which have a numerical value, i.e. only the physical parameters of a new product can be considered in the calculation.

The next stage is obtaining qualitative assessments of these parameters based on surveys of experts or customers. The assessment is given both to the researched new product and to that offered by the closest competitor. Based on the qualitative assessment given in the form 'very good’, 'good’, 'satisfactory’, 'bad’ and 'very bad’ to each of the parameters, coded scales by the number of parameters are built (customer or expert surveys). After transforming single responses into particular desirability functions, the new product competitiveness index NPGI will be obtained, as indicated in Formula (4):

where INDn represents individual needs/desirabilities.

Coded scales according to Harrington desirability function (Harrington, 1965) are as under: very good, 1.00–0.80; good, 0.80–0.63; satisfactory, 0.63–0.37; bad, 0.37–0.20; and very bad, 0.20–0.00.

The overall desirability function NPKI is defined as the geometric average of the individual desirability functions of each response INDn, where n is the number of responses. The optimal solutions are determined by maximising NPKI. This index enables the obtaining of a unified quantitative evaluation of new product competitiveness based on qualitative assessments of certain parameters. Considering that competitiveness is a multicomponent index influenced by marketing activities and other activities of companies, and new competitive advantages that appear as other groups of parameters can be used for new product competitiveness evaluation, these can then be introduced for calculating an optimised new product competitiveness index.

Coded scales according to Harrington desirability function (Harrington, 1965) are as under: very good, 1.00–0.80; good, 0.80–0.63; satisfactory, 0.63–0.37; bad, 0.37–0.20; and very bad, 0.20–0.00. The overall desirability function NPKI is defined as the geometric average of the individual desirability functions of each response INDn, where n is the number of responses. The optimal solutions are determined by maximising NPKI. This index enables the obtaining of a unified quantitative evaluation of new product competitiveness based on qualitative assessments of certain parameters. Considering that competitiveness is a multicomponent index influenced by marketing activities and other activities of companies, and new competitive advantages that appear as other groups of parameters can be used for new product competitiveness evaluation, these can then be introduced for calculating an optimised new product competitiveness index.

Conclusions

The paper presents the real market effects of new products, and their success and failure rates, from the point of view of companies representing various industries. The success or failure indicators, parameters and formal evaluation methods presented here are not expected to be exhaustive or to constitute an immediate and effective recipe for the competitiveness and success of a new product in the marketplace. The benefits of adoption of any type of measure presented in this paper depend on who is using the presented measurement methods. Nevertheless, the study does discuss immediate and pertinent issues by using established metrics for new product strategy effects’ assessment. The study provides useful metrics and methods of new product competitiveness evaluation that can be part of a more holistic and effective assessment of innovation projects. The statements made in the paper are based on examples of bibliographic sources and published empirical research on competitiveness and success or failure rates of a new product on the market. Useful research measures in the field of marketing and sales effects of new products are proposed. Simultaneously, the paper discusses direct and significant problems related to the use of established measures of new product success factors and product competitiveness. And at this point, it can be assumed that the present study’s indicators of the new product success on markets are better than those presented in the sources referenced in the paper. The research contained in the article provides indicators that can be part of a holistic and effective evaluation of new products on the competitive market.

References

1. Carboni, O. A., & Russu, P. (2018, March). Complementarity in product, process, and organizational innovation decisions: Evidence from European firms. R&D Management, 48(2), 210–222. doi:10.1111/radm.12284
2. Castellion, G., & Markham, S. K. (2013, September). Myths about new product failure rates perspective: New product failure rates: Influence of argumentum ad populum and self-interest. Journal of Product Innovation & Management, 30(5), 976–979.
doi:10.1111/j.1540-5885.2012.01009.x
3. Commission Regulation (EC) No. 1450/2004 of 13 August 2004 implementing Decision No. 1608/2003/EC of the European Parliament and of the Council concerning the production and development of Community statistics on innovation. Retrieved from https://eurlex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2004:267:0032:0035:EN:PDF
4. Cooper, R. G. (1993). Winning at new products. Reading, MA: Addison Wesley.
5. Cooper , R. G. (2002). Winning at new products. Accelerating the process from idea to lunch (3rd ed.). Reading, MA: Addison-Wesley Publishing Company.
6. Cooper, R. G. (2017). Winning at new products: Creating value through innovation (5th ed.). New York, NY: Basic Books.
7. Cooper, R. G. (2019). The drivers of success in new-product development. Industrial Marketing Management, 76, 36–47.
8. Cooper, R. G. (2021). Accelerating innovation: Some lessons from the pandemic. Journal of Product Innovation Management, 38, 221–232. doi:10.1111/jpim.12565
9. Cooper R.G., Edgett, S.J.(2008). Maximizing productivity in product innovation, in: Research Technology Management, March 1.
10. Crawford, C. M. (1979, September). New product failure rates-Facts and fallicies. Research Management, 22, 9–13.
11. Engelman, R. M., Fracasso, E. M., Schmidt, S., & Zen, A. C. (2017). Intellectual capital, absorptive capacity and product innovation. Management Decision, 55(3), 474–490. doi:10.1108/MD-05-2016-0315
12. Gabrusewicz, W. (2014). Analiza finansowa przedsiębiorstwa, Teoria i zastosowanie. Warszawa, Poland: PWE.
13. Griffin, A., & Page, A. L. (1996, November). The PDMA success measurement project: Recommended measures for product development success and failure. Journal of Product Innovation Management, 16(6), 478–496. doi:10.1111/1540-5885.1360478
14. Grzegorczyk, T., & Głowiński, R. (2020). Patent management strategies: A review. Journal of Economics & Management, 40(2), 37–39. doi:10.22367/jem. 2020.40.02
15. Harrington, E. C. Jr. (1965). The desirability function. Industrial Quality Control, 21, 494–498.
16. Heij, C. V., Volberda, H. W., Van den Bosch, F. A. J., & Hollen, R. M. A. (2020). How to leverage the impact of R&D on product innovation? The moderating effect of management innovation (March 2020). R&D Management, 50(2), 277–294. Retrieved from https://ssrn.com/abstract=3617110 or doi:10.1111/radm.12396
17. Hervas-Oliver, J. L., Sempere-Ripoll, F., Boronat-Moll, C., & Rojas-Alvarado, R. (2018). On the joint effect of technological and management innovations on performance: Increasing or diminishing returns? Technology Analysis & Strategic Management, 30, 569–581. doi:10.1080/09537325.2017.1343462
18. House, C. H., & Price, R. L. (1991). The return map: Tracking product teams. Harvard Business Review, 69(1), 92–100. Retrieved from https://hbr.org/1991/01/the-return-map-tracking-product-teams
19. Juchniewicz, M., & Grzybowska, B. (2010). Innowacyjność mikro przedsiębiorstw w Polsce. Warszawa, Poland: Polska Agencja Rozwoju Przedsiębiorczości.
20. Kasprzak, W., & Pelc, K. (2012). Innowacje, strategie techniczne i rozwojowe. Wrocław, Poland: Oficyna Wydawnicza Politechniki Wrocławskiej.
21. Kirsner, S. (2015). What big companies get wrong about innovation metrics. Retrieved from https://hbr.org/2015/05/what-big-companies-get-wrong-about-innovation-metrics
22. Lee, H., & Markham, S. K. (2016). PDMA comparative performance assessment study (CPAS): Methods and future research directions. Journal of Product Management, 33(S1), 3–19. Product Development & Management Association. doi:10.1111/jpim.12358
23. Mu, J., Thomas, E., Peng, G., & Di Benedetto, A. (2017, July). Strategic orientation and new product development performance: The role of networking capability and networking ability. Industrial Marketing Management, 64, 187–201.
24. OECD/Eurostat. (2018). Oslo manual 2018: Guidelines for collecting, reporting and using data on innovation (4th ed., The Measurement of Scientific, Technological and Innovation Activities). Paris, Luxembourg: OECD Publishing/Eurostat. doi:10.1787/ 9789264304604-en
25. Oral, M., & Kettani, O. (2009). Modelling firm competitiveness for strategy formulation. CIRRELT, 2009, 52, 1–39.
26. Reinertsen, D. G., & Smith, G. P. (2001). Developing products in half the time. New York, NY: Van Norstrand Reinhold.
27. Roostika, R., Wahyuningsih, T., & Haryono, S. (2015). The impacts of external competitiveness factors in the handicrafts industry. Polish Journal of Management Studies, 12(1), 166–176.
28. Rutkowski, I. P. (2007). Rozwój nowego produktu. Metody i uwarunkowania. Warszawa, Poland: Polskie Wydawnictwo Ekonomiczne.
29. Rutkowski, I. P. (2016). Metody innowacji produktu. Macierzowo-sieciowe metody pomiaru dojrzałości procesu innowacji produktu. Poznań, Poland: Wydawnictwo Uniwersytetu Ekonomicznego w Poznaniu.
30. Rutkowski, I. P. (2021). Competence measurement of production enterprises in product innovations for technological and marketing strategies. Journal of Economics & Management, 43, 114–134. doi:10.22367/jem.2021.43.06
31. Rutkowski, I. P. (2022). Success and failure rates of new food and non-food products introduced on the market. Journal of Marketing and Consumer Behaviour in Emerging Markets, 1(14), 52–61. doi:10.7172/2449-6634.jmcbem.2022.1.4
32. Shpak, N., Seliuchenko, N., Kharchuk, V., Kosar, N., & Sroka, W. (2019). Evaluation of product competitiveness: A case study analysis. Organizacija, 52(2), 107–125. doi:10.2478/orga-2019-0008
33. Stevens, Greg A. & Burley, James. (1997). 3,000 Raw Ideas = 1 Commercial Success!, Research-Technology Management, 40:3, 16–27, DOI: 10.1080/08956308. 1997.11671126
34. Takei, F. (1985). Product competitiveness evaluation-Quantitative analysis for development strategy. Technological Forecasting and Social Change, 28(2), 123–139. doi:10.1016/0040-1625(85)90010-1
35. Thompson, A. A. Jr., Peteraf, M. A., Gamble, J. E., & Strickland, A. J. III. (2016). Crafting and executing strategy-The quest for competitive advantage: Concepts and cases, 20th ed. New York, NY: McGraw-Hill.
36. Tyunyukova, E., Ruban, V., & Burovtsev, V. (2018). Modern approaches to product competitiveness evaluation for companies of various industries. MATEC web of conferences polytransport systems doi:10.1051/matecconf/201821602016
37. Vashkiv, O. (2020). Product competitiveness and stages of its evaluation. Figshare. Thesis. doi:10.6084/m9.figshare.13084097.v1
38. Walker, R. (2013). Winning with risk management. Hackensack, NJ: World Scientific Publishing.
39. Wheelwright, S. C., & Clark, C. B. (1992). Revolutionizing product development. New York, NY: The Free Press.
40. Wodecka-Hyjek, A. (2013). Wybrane narzędzia pomiaru innowacyjności. Zeszyty Naukowe. Zarządzanie, 998, 63–82. doi:10.15678/ZNUEK.2013.0922.04
41. Wu, D. D., Kefan, X., Gang, C., & Ping, G. (2010). A risk analysis model in concurrent engineering product development. Risk Analysis, 30(9):1440–1453.
42. Zizlavsky, O. (2016). Innovation performance measurement: Research into Czech business practice. Economic Research, 29(1), 816–838. doi:10.1080/1331677 X.2016. 1235983