Likert Scales: Always does not Exist? Maybe Never?

In the process of a multi-researcher, multinational, long-term field survey research study, we find striking differences in the use of 5-anchor never-always Likert item anchors across national samples. Two distinct types are identified, for samples using the “ALWAYS” anchor as a natural extreme of a 5=anchor set of responses, and another set of samples selecting “ALWAYS” much fewer times than “OFTEN”, indicating a unique interpretation of that anchor by study participants. No consistent relationships with cultural value dimensions have been identified using Hofstede’s seven-dimensional model, with the exception of possible relationships with the Individualism – Collectivism. Most frequently cross-cultural research studies compare mean values on constructs or in some way compare the strength of relationships between constructs across cultures. If these comparisons are to have meaning it is necessary that the data collected from respondents in different cultures validly and reliably assess the constructs being investigated, and that certain requirements of measurement equivalence be satisfied (Steenkamp & Baumgartner, 1998). Unfortunately, we have known for a long time that people`s responses to survey questions do not only measure the intended content that the items in questionnaires are trying to assess, but may also reflect content-irrelevant influences (Cronbach, 1946, 1950; Lentz, 1938). In our analyses of the distributions of the total times each anchor was selected by our national and sub-national samples, “Never” was very rarely selected and “Always” was differentially selected by some, e.g., US Hispanics and Mexicans, as a natural progression from “Often”, but by US Anglos and other high Individualism societies “Often” was selected much more frequently than “Always”. This behavior was also observed in South American and Mainland China samples. Our data indicate that the “bias” frequently stems from study participants avoiding selection of the extreme category (Always). Due to the many potential causes of this behavior further studies with interviews after completing the survey are planned.