Using letters to identify products or brands

Letters are often used as product labels in order to overcome respondent biases toward particular brands or manufacturers. However, this introduces the possibility of another source of bias, since some letters of the alphabet may be perceived more favorably than others. The results of a study conducted by Market Facts to investigate attitudes toward letters are presented. Methods of reducing possible letter bias are also discussed. It is recommended that more than one set of letter codes be used whenever possible in order to accomplish this objective.


Introduction:

When consumers are presented one or more products to evaluate, their responses may be affected by awareness of the brand names or manufacturers of those products. Thus, a respondent may rate a particular product higher or lower if given the brand name than would be the case if the brand name was unavailable. This could occur, for example, because one of the products rated is the respondent's "usual" brand or because the respondent had a bad experience with a particular product. Obviously, such respondent biases can pose problems in product comparison studies.

To overcome such bias, letters rather than brand names are often used to label the products presented to respondents. Presumably, the individuals then must base their evaluations on the characteristics of the products themselves, rather than on existing attitudes toward the brand or manufacturer. However, this introduces the possibility of a different source of bias: preference for certain letters. If a respondent truly has no preference for one product over the other(s), is (s)he likely to choose Brand "A" over Brands "B" and "C" simply because of a preference for that letter?

Respondent bias toward letters is probably less serious than bias toward most actual brand names. Nevertheless, it is important to determine the extent to which systematic tendencies exist for respondents to favor products identified by particular letters. This report presents data from a preliminary study carried out by Market Facts to determine whether some letters are perceived more favorably than others.


Method:

Questionnaires were mailed to 4,000 households drawn from the Market Facts' Consumer Mail Panel. Each sub-sample of 1000 was balanced to be nationally representative with respect to geographic region, population density, age and income. Approximately 3,000 questionnaires were returned.

The instructions given to the respondents were as follows:

The numbers in parentheses are the values assigned to the responses for the analysis of the data (and were not shown to the respondents.)


Results:

The mean ratings of each letter for the total sample and for each version are shown in Table One. In each column, the letters are listed in order of the mean rating. The mean ratings for the total sample are also depicted graphically in Figure One, with the letters arranged in alphabetical order.

There is a great deal of consistency in the rank order of the letters, although some minor differences exist among the versions. In particular, the top third of the letters are the same in all versions, as are the bottom third. Thus, it appears that the order in which the letters were presented had little effect on the ratings.

With a few notable exceptions, letters near the beginning of the alphabet tended to be rated higher than those near the end. Letters A and B were rated the highest in all versions and letters U through Z were among the lowest. However, the letters M and S (and, to a lesser extent, L, R, and T) were rated high and letters F and Q were rated low in all versions.


Discussion:

Since some letters were consistently perceived as more favorable than others, care should be exercised in the selection of letters as brand labels. Choosing letters which are relatively similar with respect to their ratings may help to minimize bias caused by the product labels. Letters G thru P (except L and M) were relatively homogeneous in their ratings. Of course, the letters used should not suggest or appear to be abbreviations for the actual brand names.

It is important to recognize that the differences among letters found in this study may be less extreme when letters are actually used to label products. In this study, respondents were instructed to think specifically about the letters themselves. In most situations, however, respondents are asked to focus on the products rather than on the designations or labels. When the letters are secondary to the respondent's task, the biases are likely to be smaller than reported here. Market Facts is engaged in further research to assess the magnitude of letter bias when the letters are actually used as product labels.


Recommendations:

Perhaps the best way to minimize letter bias is to avoid using the same product letter codes for all respondents. For example, if a study were carried out in which respondents evaluated two products, half of the respondents could see product 1 labeled as "G" and product 2 as "H," while the other respondents could see product 1 as "H" and product 2 as "G." In this case, since each brand is labeled with the letter G half the time and with the letter H half the time, any bias due to the letters would be expected to be averaged out.

Another coding scheme which could be used in this case is to label product 1 as "G1" and product 2 as "H1" for half of the respondents and to label product 1 as "H2" and product 2 as "G2" for the other respondents. Again, each product is labeled with the letter G half the time and with the letter H half the time. Since a respondent would see the same subscript in both product codes, that digit should not cause any bias.

The latter coding scheme has the advantage of being easily generalized to any number of products. For example, if four products are shown to each respondent, the coding scheme in Table Two could be used. A fourth of the respondents would see the first set of labels, a fourth would see the second set, etc.

Table Two
Possible sets of labels for four products

  Labels
Product Set 1 Set 2 Set 3 Set 4
1 G 1 H 2 I 3 J 4
2 H 1 I 2 J 3 G 4
3 I 1 J 2 G 3 H 4
4 J 1 G 2 H 3 I 4


If there are as many sets of letter codes as there are products, it is possible to use each letter in the code for each product, as in the preceding examples. This should minimize the degree of letter bias. In order to reduce costs, fewer sets of letter codes could be used. Although one would not expect the letter bias to be averaged out in this case, this goal could be at least partially achieved. Therefore, whenever letters are used to identify products, it is recommended that more than one set of identifications be employed.

If products are sequentially presented to each respondent, one can avoid the product identification problem entirely. In this case, the products can be referred to as simply "used first," "used second," etc. or as "first product," "second product," etc. Of course, it is important that the order of presentation of the products be either (a) randomized for each respondent or (b) balanced across respondents to eliminate bias due to the order of presentation.


Click to enlarge




Click to enlarge


Computer Generated Graphics provide:

Ease of interpretation
Quick visual analysis

The bar chart in Figure One is just one of the many computer generated graphic capabilities available from Market Facts. Others include: