Numerous interpretations of empathy have emphasized its importance in the field of psychology, as well as its positive influence on an individual’s growth in moral development (Jolliffee & Farrington, 2006). Hogan (1969) has particularly noted these points, claiming that empathy is an abstract construct that can be viewed as an ability to hold what he calls a “moral point of view” (p. 307). In other words, it can be seen as a willingness to step into another person’s shoes, to regard that person’s current viewpoint, and to take action in a way that is mindful of that person’s well being (Hogan, 1969). While empathy is also seen as being affective, Hogan’s (1969) analysis of the concept is primarily cognitive in nature (Jolliffe & Farrington, 2006, Carre´, Stefaniak, D’ambrosio and Bensalah, and Besche-Richard, 2013). Hogan (1969) defined this concept as, “the intellectual or imaginative apprehension of another’s condition or state of mind” (p. 307). Among the various tests and measurements that have been created in order to assess empathy, Hogan’s (1969) own empathy scale is one of them.
Developed in 1969, the Hogan Empathy Scale (EM) is a 64-item questionnaire answered in terms of true or false responses. This self-report measure is comprised of a number of items predominantly pulled from the California Psychological Inventory (CPI) and the Minnesota Multiphasic Personality Inventory (MMPI), based on assessments done at the University at California’s Institute of Personality Assessment and Research (IPAR) (Hogan, 1969). After establishing the empathy criterion via descriptions of “a highly empathetic man” (Hogan, 1969, p. 309) from faculty and students of psychology, two groups of IPAR samples embodying military officers, research scientists and student engineers had their empathy evaluated to this criterion (Hogan 1969). Sorted into high and low empathy groups based on their ratings, the final scale was created with CPI and MMPI items reflecting the highest levels of discriminability (Froman and Peloquin, 2001). This was done after both subscales were compared to the responses of the two aforementioned groups (Hogan 1969). Following the completion of the EM, Hogan (1969) found the scale’s validity and reliability to be adequate, yet, at the time, even he admits in his article that additional evaluation is needed to further reinforce his findings for practical and theoretical purposes.
In determining its practicality, the EM has been applied to various situations. Such practicality is not restricted to psychologists. This is supported by Hogan’s (1969) finding that empathy, with regards to the concept’s behavioral aspects, is construed similarly among people who have no professional or academic affiliation with psychology. In their study, Froman and Peloquin (2001) assessed the scale by having 300 occupational therapy students complete it. In this case, they described similar characteristics to empathy that a student is expected to acquire for the profession. Additionally, Johnson, Cheek, and Smither (1983) explain various situations in which the EM has been used. This includes Daurio’s finding (as cited in Johnson, Cheek, and Smither, 1983) that the scale was a good predictor of sociopolitical intelligence, and Kendall and Wilcox’s finding (as cited in Johnson et al., 1983) of empathy scores significantly correlating to therapy effectiveness in terms of improvements in children being treated. Here, Kendall and Wilcox (as cited in Johnson et al., 1983) note the stability of their study’s high empathy scores with the psychology student’s empathy scores in Hogan’s (1969) development of the EM. Furthermore, the 39-item version of the EM was used to measure cognitive aspects of empathy in offenders versus non-offenders, along with affective aspects of the construct (Jolliffee & Farrington, 2004).
Regarding the scale’s validity, Hogan (1969) found there to be adequate results, as previously mentioned. For instance, as part of determining whether or not the scale was designed to measure the construct of empathy, concurrent validity is assessed by comparing the EM to ratings of social acuity via correlation coefficients, which is a statistic generally used for this validity type (Hogan 1969). If concurrent validity is to be high, the EM should have a strong relationship with social acuity in the sample being rated simultaneously, as social acuity was deemed as being representative of empathy. After assessment staff rated the social acuity of IPAR sample groups, both scales correlated with an average coefficient of .58 (Hogan 1969). While acknowledging that the correlation coefficient was quite high due to the fact that a majority of the same staff also rated the same IPAR samples to the empathy criterion in the creation of the EM, another sample of medical school applicants were used to verify the concurrent validity being measured (Hogan 1969). Here, the correlation coefficient was significantly lower between the two scales at .42 (Hogan 1969). As validity coefficients of .30-.40 are considered pretty well for relevance of a test to the criterion, it seems that this is the case for the EM and social acuity.
Froman and Peloquin (2001) evaluated the validity of the EM as well. More specifically, convergent validity was determined through the use of a correlation coefficient, which is the statistic used for this type of validity. This level of validity and statistic was appropriately used, as the construct of empathy that the EM is supposed to be measuring was compared to variables of the Fieldwork Performance Evaluation (FWPE), which are deemed to be most closely related to that same construct. By doing this, such validity evidence was gathered to determine whether or not the EM is in fact measuring what it is supposed to be measuring, as Hogan (1969) had used the aforementioned measurement of concurrent validity for.
The FWPE is an assessment of the necessary skills acquired by students for the practice of occupational therapy (Froman and Peloquin, 2001). The scores of the FWPE are meant to evaluate interpersonal skills, relationships, and psychosocial attitudes (Froman and Peloquin, 2001). Such variables are seen as conceptually similar to the construct of empathy measured by the EM. Consequently, it is initially predicted in the study that scores of the FWPE, specifically that of the Attitude scale and treatment sub score of the Performance scale, would present convergent validity with EM scores. However, after being calculated from each of the data of both time points in the study, correlations for the Attitude scale come out to be .19 and .07 with the EM, and treatment sub score of the Performance scale come out to be .16 and .01 with the EM (Froman and Peloquin, 2001). The ideal correlation coefficient, for a positive relationship, would be as close to 1.0 as possible, or much higher than the aforementioned values. Thus, these given correlations are very low, and EM scores are not found to have a meaningful relationship with FWPE scores in this regard.
While convergent validity is being appropriately used to see whether the construct being measured by the EM relates to similar variables of the FWPE, the results do not indicate the ideal correlation coefficient value for the study at hand. In these instances, it is important to note that it can often be hard to pinpoint the source of the issue, or error, contributing to the low correlations. It could have to do with the fact that the variables of the FWPE are not measuring, or are not truly related to, the construct of empathy. It could alternatively be that both tests being compared in the study are flawed when it comes to appropriately measuring the construct of empathy. In this case, however, Froman and Peloquin (2001) suggest that only the EM may be flawed due to poor measurement, cautioning those who want to use it for long-term projects such as the one discussed here.
In addition to assessing the EM’s convergent validity, Froman and Peloquin (2001) evaluated the scale’s discriminant validity. Here, correlational analysis is also employed. This level of validity and statistic is appropriately used, as empathy, in this case considered from a mainly cognitive perspective, is compared to variables of age, gender and GPA, which are viewed to be unrelated to the construct at hand (Froman and Peloquin, 2001). By doing this, it can be shown that the EM only measures empathy, and not other variables that the former is supposed to be unassociated with.
EM scores were correlated with each of the aforementioned variables at both time points presented in the study via pairwise correlations for junior and senior year (Froman and Peloquin, 2001). Correlations that resulted were low, varying between .01 and .13, and small percentages of shared deviance were seen as well (i.e. within the range of 0% and 1.7%). As low values are ideal for discriminant validity, these indicate that EM scores are measuring a construct that has nothing to do with the biological variables mentioned, as should conceptually be the case. Thus, while Froman and Peloquin (2001) find there to be weak convergent validity, in that the EM may not be measuring what it says it is, they also find there to be strong discriminant validity, indicating that the EM may be measuring a construct that is unique and independent of other variables, with the former seemingly relevant to the scale at hand. In this case, such discriminant validity may indicate that, while the scale may not be fully assessing empathy, it could be measuring a construct closely similar to it.
It is important to note the fact that while Froman and Peloquin (2001) mention that the EM assumes empathy to be “unidimensional” (Froman and Peloquin, 2001, p. 561) in nature, Davis (1983), in contrast, finds the construct to be multidimensional. Having used Hogan’s (1969) scale as evidence, Davis (1983) utilizes concurrent validity by obtaining correlation coefficients of the EM to the four subscales of his own test of empathy, the Interpersonal Reactivity Index (IRI). The subscale of Perspective-Taking was found to correlate the best with the EM, with a value of .42 (Davis, 1983). This, then, highlights the cognitive aspect of empathy, whereas the EM correlates lower to the emotional aspects of the construct measured by the remaining three subscales of Fantasy, Empathetic Concern, and Personal Distress (Davis, 1983). The first was valued at .15, the second at .11, and the third at -.25 (Davis, 1983). If empathy is a multidimensional concept, perhaps that could account for Froman and Peloquin’s (2001) poor convergent validity and better discriminant validity of the EM. Perhaps the EM does not relate to the FWPE because it is in fact measuring an aspect of empathy that variables of the FWPE is not relevant to, which the discriminant validity value could explain for.
A test cannot be valid unless it is reliable. Regarding the scale’s reliability, Hogan (1969) found there to be satisfactory results. After having developed the scale, Hogan (1969) utilizes test-retest reliability to assess how consistent scores of the EM are overtime with correlation coefficients. Having administered the test twice, once before and once after 2 months, to 50 college undergraduate students, final correlations came out to be at a high of .84 (Hogan, 1969). However, it is important to note that such a high correlation could be due to a number of factors. Memory effects could have affected the value, as perhaps these students were able to memorize what they had put down the first time, and thus, decided to stick with the answers that they had chosen previously.
In their assessment of the EM, Froman and Peloquin (2001) test the scale’s internal consistency through the use of Cronbach’s coefficient alpha, a statistic often used to indicate the former. This level of statistic is appropriate for the length of the test (64-items) and its format (true or false responses). In other words, because of the test’s dichotomous nature, Cronbach’s alpha can be calculated due to its ability to measure such continuous variables. However, it is important to note that long scale testing, such as the EM, can be prone to having a poor relation of items, and, as a result, the alpha can be used to make the internal consistency look better than it actually is. Furthermore, a long-scale test can have a redundancy in items, making the alpha look high, when it is really representing a small portion of the construct at hand.
Despite the aforementioned possibilities, it seems that none of this is the case in the article, as the reliability of the EM is found to be poor, with alpha estimates measuring below higher ones that were reported in previous studies. For instance, in their assessment of the EM scale, Johnson et al. (1983) found the internal consistency reliability to be .69 in their study, which is close to Hogan’s (1969) internal consistency value of .71 that he obtained by using the the KR-21 formula. It is important to note that while Johnson et al. (1983) measured the scale in Liker format, equal factor structure was found with the test’s dichotomous format, making findings relevant to the case at hand. For Froman and Peloquin (2001), their internal consistency value was found to be .57, showing that, in their study, this does not replicate the aforementioned past reliability estimates of the test measured, putting into question whether the various items of the test measure the same construct of empathy.
In their study of 95 subjects between the ages of 19 to 63 years, Cross and Sharpley (1982) also found poor reliability results regarding internal consistency. Also using Cronbach’s alpha, such a value came out to be .605 (Cross and Sharpley, 1982). In this case, this is undesirable, as acceptable alpha scores, which have no clear consensus, should be a minimum of .70, .80, or .90. To understand why this may be the case, Cross and Sharpley (1982) did an additional in-depth analysis, which not only accounted for redundancy in a majority of the items, but also characterized those items as contradictory to the EM’s interpretation of empathy. Furthermore, about half of the items were irrelevant to the total score of the test (Cross and Sharpley, 1982). As mentioned previously, the length of the test could account for the redundancy of the items, although it seems even this does not lead to a very high alpha value, due to the irrelevancy of the items to the scale. Cross and Sharpley (1982) contend that this could be due to Hogan (1969) having created the scale with the items from the MMPI and the CPI, which may not be good for the EM’s homogeneity.
Hogan (1969) intended to have a scale that measured the construct of empathy as a whole. Yet, this may not be the case based on the aforementioned assessments of the EM’s validity and reliability. However it is important to note that the EM does have its strengths. As mentioned previously, EM’s practicality is flexible, as it can be administered to men and women of different ages for a variety of purposes. The EM’s dichotomous format can also be efficient for its length, by having individuals respond quickly to each item using their absolute judgment. Additionally, Hogan (1969) provides evidence that test scores remain stable overtime, which empathy, as ability or trait, can be characterized as.
However, even with its strengths, the test still has serious limitations. The dichotomous format has its flaws, as it is hard to account for complexity when it comes to measuring the intended construct. Even if Johnson, Cheek, and Smither (1983) used a Likert format, a majority of the studies utilized the EM via true or false responses. Also, if the EM is measuring the cognitive aspect of empathy, this goes against the test’s assumption of empathy as being defined as unidimensional. This could account for the contradictory nature found by Cross and Sharpley (1982) regarding the items relation to the total test score with the poor internal consistency found.
It is important to note that there seems to be poorer results regarding reliability, which could be weakening the scale’s construct validity as a whole, based on the previous information provided. With this in mind, the EM may not be as suitable when it comes to serving its desired purposes. However, as finding the validity of a test is an ongoing process, and there is a limited amount of assessments done on the EM’s psychometric properties, further evaluations of the test should be considered.
