1 00:00:10,042 --> 00:00:16,014 Mark. You know this. Is a bit of 2 00:00:16,015 --> 00:00:23,005 a mouthful. All right so we're going to talk about 3 00:00:23,006 --> 00:00:28,055 the nest our personality disorders and I'm going to use an ira key analysis as I 4 00:00:28,056 --> 00:00:34,060 mentioned phases and we're going to look at ethnic or diversity within the results 5 00:00:34,060 --> 00:00:41,058 . All right so before we dive into it 6 00:00:41,095 --> 00:00:48,022 is it Ok to use personality and measure with minority populations. We've kind of 7 00:00:48,023 --> 00:00:49,097 been taught in school that you know if something is 8 00:00:49,098 --> 00:00:55,086 a representative sample that it may or may not be Ok in some situations right so in 9 00:00:55,087 --> 00:01:00,093 order to delve into a little bit deeper we're going to look at 1st the d.s.m. 10 00:01:00,097 --> 00:01:06,022 And then we're going to go into more of the actual data behind the d.s.m. 11 00:01:06,058 --> 00:01:08,097 Warning. So the d.s.m. 12 00:01:08,098 --> 00:01:09,051 Indicates that 13 00:01:09,052 --> 00:01:16,047 a personality disorder. Faces inflexible and certain pattern of experiences that 14 00:01:16,048 --> 00:01:20,051 deviates markedly from societal expectations and causes distrust and impairment 15 00:01:20,052 --> 00:01:23,071 right we all know this because the d.s.m. 16 00:01:23,072 --> 00:01:30,040 Warns clinicians not over under diagnosis based on minority status and in fact in 17 00:01:30,058 --> 00:01:34,066 clude statement that judgments must take into account the individual's cultural 18 00:01:34,067 --> 00:01:39,045 background and personality disorder should not be confused with problems associated 19 00:01:39,046 --> 00:01:44,006 with acculturation customs or religious values and includes 20 00:01:44,007 --> 00:01:50,100 a whole bunch of warnings in the in the culture considerations section. I have them 21 00:01:51,001 --> 00:01:52,025 all up there I'm just going to give you 22 00:01:52,026 --> 00:01:56,066 a brief highlight of some of them for example to get the Temple says that rituals 23 00:01:56,067 --> 00:02:02,007 such as who do life beyond death shaman ism the evil eye magical beliefs related to 24 00:02:02,008 --> 00:02:07,004 health and illness should be taken into account for histrionics for example it says 25 00:02:07,042 --> 00:02:11,057 interpersonal behaviors emotional expression and physical appearance vary across 26 00:02:11,058 --> 00:02:17,002 cultures and the narcissistic personality disorder actually did not include any 27 00:02:17,003 --> 00:02:24,002 warnings whatsoever All right so research to 28 00:02:24,003 --> 00:02:29,000 this day has not included very much most of the stuff that has been completed so 29 00:02:29,001 --> 00:02:30,041 far it's been on the m.m.p.i. 30 00:02:30,047 --> 00:02:36,089 Rather than other measures. A few results that were notable for Native Americans 31 00:02:36,090 --> 00:02:42,083 there was 3 studies completed they had very similar profiles with elevations on 32 00:02:42,084 --> 00:02:49,073 scales f 1468 and 9 appearing most often in fact some of them included 33 00:02:50,057 --> 00:02:56,035 all of the scales being. Elevated for some and having 34 00:02:56,036 --> 00:03:02,005 a Over 50 percent misdiagnosis rate for African-Americans it was during result the 35 00:03:02,006 --> 00:03:09,000 most common was. 8 and 9. They did have differences and up to 39 percent of the 36 00:03:09,001 --> 00:03:15,060 items so in 1920 decided they. Should not be 37 00:03:15,095 --> 00:03:20,069 used with minorities at all since we don't know how the ethnic differences can be 38 00:03:20,070 --> 00:03:27,068 applied to interpretation All right so why do we care about the nest 39 00:03:27,069 --> 00:03:31,092 our data. The end I triple a completed 40 00:03:31,093 --> 00:03:38,043 a nationwide study 3 waves completed in 2001 all the way to 41 00:03:38,044 --> 00:03:41,010 2013 they use the oddities for which is 42 00:03:41,011 --> 00:03:46,012 a structured interview of alcohol related behaviors consisting of yes or no 43 00:03:46,031 --> 00:03:48,038 questions. To receive 44 00:03:48,039 --> 00:03:52,048 a diagnosis participants had to endorse at least one symptom causing significant 45 00:03:52,049 --> 00:03:58,057 dysfunction above and beyond a reckless a number of items so the d.s.m. 46 00:03:58,061 --> 00:03:59,076 5 includes an s. 47 00:03:59,077 --> 00:04:05,030 Our data for its prevalence to test 6 this is why we care. So it's really important 48 00:04:05,031 --> 00:04:09,090 to ensure that the Nets our data is not biased towards minorities. The next our 49 00:04:09,091 --> 00:04:14,052 data suggests that approximately 15 percent of us adults will have at least one 50 00:04:14,053 --> 00:04:20,002 personality disorder and no questions on the status that actually includes whether 51 00:04:20,003 --> 00:04:21,004 or not that behavior is 52 00:04:21,005 --> 00:04:25,031 a deviation from the participants culture or whether or not it's maladaptive within 53 00:04:25,032 --> 00:04:30,085 their culture. All right so for the present study I had 54 00:04:30,086 --> 00:04:36,026 a couple of hypotheses 1st was that collectivist cultures such as Asian native 55 00:04:36,027 --> 00:04:42,004 Hispanics would be more likely to endorse items on the dependent scale. That Native 56 00:04:42,005 --> 00:04:46,076 Americans would be more likely to endorse items related to spiritual or magical 57 00:04:46,077 --> 00:04:53,047 ideas. This is the item that tips fail and that African-Americans would be more 58 00:04:53,048 --> 00:04:58,022 likely to endorse items of persecution which are present on the paranoid scale 59 00:05:00,063 --> 00:05:06,042 alright so far hasn't been swayed one has $43003.00 participants from across the 60 00:05:06,043 --> 00:05:11,029 United States of non institutionalized people 18 or older that include military 61 00:05:11,030 --> 00:05:14,084 service. The data was adjusted for 62 00:05:14,085 --> 00:05:21,022 a socio demographic factors at the time and primary recognition of minority status 63 00:05:21,023 --> 00:05:25,030 was used in this because people could indicate more than one minority status 64 00:05:26,004 --> 00:05:30,085 approximately 20 percent were African-American 100 percent Hispanic or Latino 3 65 00:05:30,086 --> 00:05:37,085 percent Native American 3 percent Asian and roughly 57 percent were female so for 66 00:05:37,086 --> 00:05:44,038 Wave 2 they had an 86.7 percent response rate resulting in $34653.00 67 00:05:44,039 --> 00:05:50,051 participants from the way one study. The study was conducted in the participants 68 00:05:50,052 --> 00:05:57,045 home and. They completed 7 personality measures for the 1st scale and 4 69 00:05:57,046 --> 00:06:04,037 for the 2nd. The anti social was completed twice on so for both scales but the way 70 00:06:04,038 --> 00:06:10,079 one data was used because it was more because and way 3 wasn't used ironically 71 00:06:10,080 --> 00:06:11,054 because they use 72 00:06:11,055 --> 00:06:15,023 a different scale and they had different participants so I didn't want to 73 00:06:15,038 --> 00:06:20,085 complicate things. So I missed in this I don't response theory what is it. 74 00:06:22,080 --> 00:06:28,093 At its core it looks at observed differences in responses between groups and 75 00:06:28,094 --> 00:06:32,087 checking to see if the item or test is biased or to see whether or not there's true 76 00:06:32,088 --> 00:06:38,064 differences between the population or both. It has the ability to distinguish 77 00:06:38,065 --> 00:06:43,045 between these groups differences and the bias so at its core I or t. 78 00:06:43,049 --> 00:06:49,012 Will. Compare to regression slope as you can see as an example of it over on the 79 00:06:49,013 --> 00:06:56,006 corner and it asks an essential question so given the person score on an underlying 80 00:06:56,007 --> 00:07:00,091 construct or trait was that person's true score when they're viewed as 81 00:07:00,092 --> 00:07:03,079 a member of the focus group and what if they were viewed as 82 00:07:03,080 --> 00:07:05,080 a member of the reference group so sometimes it's 83 00:07:05,081 --> 00:07:12,078 a change and that showing that there's bias in the groups so for besiegers the 1st 84 00:07:12,079 --> 00:07:13,039 thing to do with 85 00:07:13,040 --> 00:07:16,060 a complete unit dimensionally check because that's one of the assumptions of an 86 00:07:16,061 --> 00:07:16,095 r.t. 87 00:07:16,096 --> 00:07:23,075 Analysis. However I argue analysis is really robust to all of its. Assumptions 88 00:07:23,076 --> 00:07:25,049 and there are other models of i.r.t. 89 00:07:25,050 --> 00:07:30,027 That exist for when those assumptions are broken. So the 1st one completed was 90 00:07:30,028 --> 00:07:32,042 a modified parallel analysis and 91 00:07:32,043 --> 00:07:37,027 a maximum likelihood ratio of those indicated multi-dimensionality for over half 92 00:07:37,028 --> 00:07:42,052 the scales this indicated that the one parameter model was not appropriate and was 93 00:07:42,053 --> 00:07:49,011 also really expected because both of those use. The use 94 00:07:49,012 --> 00:07:55,055 significance I use and given the large data set I was kind of expecting that. So in 95 00:07:55,056 --> 00:08:00,066 order to determine what was really going on I compared the eigenvalues of 1st and 96 00:08:00,067 --> 00:08:06,014 2nd eigenvalues and indicated that there was one dominant factor for all of the 97 00:08:06,015 --> 00:08:10,067 groups except for Native Americans on the anti social scale which was extremely 98 00:08:10,068 --> 00:08:16,038 close to the cutoff So overall I felt that 2 parameter model would have been most 99 00:08:16,039 --> 00:08:21,035 appropriate and felt that since there was generally one dominant factor that it 100 00:08:21,036 --> 00:08:28,008 would be Ok to go ahead with the analyses. So. For the analyses there were 101 00:08:28,009 --> 00:08:30,084 2 options available a mental handle and 102 00:08:30,085 --> 00:08:34,046 a legislature question or approach I will not get into the differences but 103 00:08:35,032 --> 00:08:41,010 basically both of them use kies wares and are prone to type one error on the ledge 104 00:08:41,011 --> 00:08:45,014 a succession approach has a bit lower power which can balance it out 105 00:08:45,015 --> 00:08:50,031 a little bit in larger sample so I was using that as my primary thing and I also 106 00:08:50,032 --> 00:08:56,055 calculated odds ratios to examined the magnitude of effect or the impact of the 107 00:08:56,056 --> 00:09:03,037 potential bias. All right so one of the joys things 108 00:09:03,038 --> 00:09:07,007 about this was I got the actual prevalence rates for everyone so across 109 00:09:07,008 --> 00:09:12,096 participants obsessive compulsive had the highest prevalence at 7.8 percent and 110 00:09:12,097 --> 00:09:19,068 dependent had the lowest at proximately point 5. Data was not available for Native 111 00:09:19,069 --> 00:09:24,090 American and Asians for the dependent scale this was because there was 15 percent 112 00:09:25,010 --> 00:09:29,091 so there was only 15 precious fence or less endorsing the disorder the data is 113 00:09:29,095 --> 00:09:35,073 controlled by census and they will not allow for privacy sake information with less 114 00:09:35,074 --> 00:09:42,074 than 15 participants available for publication so Asian Americans 115 00:09:42,075 --> 00:09:43,087 had almost half 116 00:09:43,088 --> 00:09:49,087 a probability for over half of the disorders. In contrast African-Americans are 117 00:09:49,088 --> 00:09:54,099 twice as likely to have paranoid schizo typical and narcissistic personality 118 00:09:54,100 --> 00:10:01,003 disorders and Native Americans were 2 to 3 times likely to have schizotypal 119 00:10:01,004 --> 00:10:07,003 paranoid. Skit so I ordered mine and 3 times likely for anti social. 120 00:10:10,084 --> 00:10:11,045 All right I have 121 00:10:11,046 --> 00:10:16,093 a lot of things that came out of significant which is really terrifying this is 122 00:10:16,094 --> 00:10:20,062 just a summary of it. Overall there were such 123 00:10:20,066 --> 00:10:23,094 a large number that there was no physical way that I could possibly get through it 124 00:10:23,098 --> 00:10:26,014 in a day talking to you so 125 00:10:26,044 --> 00:10:31,060 a couple of highlights 17 out of 18 items for African-Americans scale. 126 00:10:33,008 --> 00:10:33,088 Bias for 127 00:10:33,089 --> 00:10:39,038 a personality for the narcissistic personality disorder and Asians actually in 128 00:10:39,039 --> 00:10:45,086 about almost half of the sales had none so 129 00:10:46,039 --> 00:10:50,090 obviously I differential I don't functioning on these skills could be contributing 130 00:10:50,091 --> 00:10:57,082 at this scale or Test level right this is an awful lot of bias going on so again I 131 00:10:57,083 --> 00:11:02,053 have so many results from so many different things that it was terrifying that I 132 00:11:02,054 --> 00:11:09,032 can only give you the highlights so at the scale level. You have bias that had 133 00:11:09,033 --> 00:11:15,075 notable impact and possible was that one logistic regression or the man who has 134 00:11:15,076 --> 00:11:22,061 a coming out significant but it may not have had notable odds ratio so I'm 135 00:11:22,062 --> 00:11:24,040 only going to talk about the ones that did have 136 00:11:24,041 --> 00:11:30,100 a notable impact today so for paranoid personality disorder African-Americans were 137 00:11:31,001 --> 00:11:33,017 noted as having bias on the scale with 138 00:11:33,021 --> 00:11:39,084 a significant point 58 to. Skitt So I personally this order 139 00:11:39,085 --> 00:11:46,073 was notable for African-Americans schizotypal had biased towards 140 00:11:46,074 --> 00:11:51,100 African-Americans and Native Americans they had an entree show points. 141 00:11:54,005 --> 00:11:55,098 In this one they actually did not have 142 00:11:55,099 --> 00:12:01,029 a significant mental handle which is unusual. For borderline personality disorder 143 00:12:01,047 --> 00:12:07,088 women came out as having significant bias for narcissistic also women which was 144 00:12:07,089 --> 00:12:14,076 a little bit surprising Hispanics and African. For any social 145 00:12:15,018 --> 00:12:16,024 Ironically there was 146 00:12:16,025 --> 00:12:21,003 a significant logistic regression for all of the groups except for Native Americans 147 00:12:21,049 --> 00:12:27,093 but only 2 of them had high impact which was for males and Hispanics for histrionic 148 00:12:27,094 --> 00:12:34,075 disorder personality disorder African-Americans and. Bias. With odds 149 00:12:34,076 --> 00:12:41,052 ratio point 553. However good news avoidant obsessive compulsive and 150 00:12:41,056 --> 00:12:46,036 dependent personality disorders did not have any significant logistic regression or 151 00:12:46,063 --> 00:12:51,096 mental handle crisis where However the dependence Caleb's noted that we didn't have 152 00:12:51,097 --> 00:12:52,084 information for 153 00:12:53,029 --> 00:12:58,076 a native and Asian All right so 154 00:12:59,070 --> 00:13:05,042 a lot of information what does all that mean. So differential I don't function was 155 00:13:05,043 --> 00:13:10,087 obviously found throughout the next arc data in most cases the items actually did 156 00:13:10,088 --> 00:13:11,076 not cause 157 00:13:11,077 --> 00:13:17,024 a significant concern as bias was cancelled out at the Test level or it was not 158 00:13:17,025 --> 00:13:23,022 very impactful overall. The paranoid and schizotypal items bias were more impactful 159 00:13:23,023 --> 00:13:27,098 than I originally expected in my how posses was bias at the Test level which I was 160 00:13:27,099 --> 00:13:33,033 not actually anticipating I was very optimistic. African-Americans throughout the 161 00:13:33,034 --> 00:13:37,061 next arc was surprising and kind of terrifying with bias on over half the scale. 162 00:13:40,039 --> 00:13:45,071 In the end personality disorder should really be interpreted with extreme caution 163 00:13:45,084 --> 00:13:45,099 for 164 00:13:46,000 --> 00:13:50,099 a minority populations and clinicians the to be aware of spiritual beliefs cultural 165 00:13:51,000 --> 00:13:57,020 differences when completing diagnoses. Without this many minority clients might be 166 00:13:57,021 --> 00:13:58,083 assumed to meet criteria for 167 00:13:58,084 --> 00:14:05,069 a personality disorder when it's not really otherwise indicated Does anyone have an 168 00:14:05,069 --> 00:14:07,088 . I know I used it for 169 00:14:07,089 --> 00:14:23,033 a lot of that information really quickly. Yes. The 170 00:14:23,034 --> 00:14:29,025 way that census coded the data was actually for the odds ratios were actually 171 00:14:29,026 --> 00:14:35,032 calculated for Caucasians compared to the minorities class so in the end result was 172 00:14:35,033 --> 00:14:35,092 that if it was 173 00:14:35,093 --> 00:14:42,081 a point 5 That means that the Caucasians were actually almost half as less likely 174 00:14:42,082 --> 00:14:49,050 to have that odds ratio so in my actual data I left them around so that I could say 175 00:14:49,051 --> 00:14:54,007 that it was like say they were twice as likely to have it as opposed to Caucasians 176 00:14:54,008 --> 00:14:55,065 being half as likely so it made 177 00:14:55,066 --> 00:14:59,095 a little bit more sense but the original odds ratios that were presented were 178 00:15:00,083 --> 00:15:03,080 reversed from what you would naturally think of when you're comparing them. 179 00:15:14,090 --> 00:15:18,025 I didn't have a chance to do a lot of that there has been 180 00:15:18,026 --> 00:15:24,080 a lot of data analyses completed on this in terms of the validity 181 00:15:24,081 --> 00:15:30,002 and reliability of the measures and the participants and everything else 182 00:15:30,003 --> 00:15:34,022 a lot of that has actually been really completed by other people when the analysis 183 00:15:34,023 --> 00:15:41,018 was originally done by Bridget Grant and her team. But I 184 00:15:41,019 --> 00:15:47,038 didn't actually do any of that myself census really controls how they use the data 185 00:15:47,039 --> 00:15:52,008 nowadays. Back when the original way one little time you could just get the 186 00:15:52,009 --> 00:15:55,003 original data set and do whatever you wanted with it but now you have to make 187 00:15:55,004 --> 00:16:01,100 requests to census to have someone over there completed all for you so really was 188 00:16:02,001 --> 00:16:05,048 kind of limited on what I could do so you had 189 00:16:05,049 --> 00:16:13,006 a question. The 190 00:16:13,007 --> 00:16:17,039 one from better model releases assumes that there is no influence whatsoever from 191 00:16:17,040 --> 00:16:23,015 any other factor so we're assuming here that it's not just their responses that 192 00:16:23,016 --> 00:16:26,074 we're assuming that there's more than one thing going on like it's not just the 193 00:16:26,075 --> 00:16:31,004 person there's race involved so basically any time that you add in 194 00:16:31,038 --> 00:16:36,054 a dimension you'd be needing to change your premise or model it goes up to 3 which 195 00:16:36,058 --> 00:16:36,074 is 196 00:16:36,075 --> 00:16:42,035 a multi-dimensional model so one parameter model is really limited no one can and cannot 197 00:16:42,036 --> 00:16:45,042 do and what you can and cannot assume the 2 premier model allows 198 00:16:45,043 --> 00:16:46,084 a little bit more comparison and. 199 00:16:52,041 --> 00:16:56,080 That was what was voted in the original. 200 00:16:58,100 --> 00:17:05,081 Request that I provided for them. When I completed it. Originally. 201 00:17:07,029 --> 00:17:13,062 I coded it out for the race being the one that was being modified and it does what 202 00:17:13,063 --> 00:17:17,050 the group membership was based on and that's usually the 2nd parameter otherwise I 203 00:17:17,051 --> 00:17:20,060 would have been able to have coded for the group in there because it would have 204 00:17:20,061 --> 00:17:21,090 just shoved everyone in 205 00:17:29,099 --> 00:17:36,080 it is entirely possible however. In order to do that I would have had to 206 00:17:36,081 --> 00:17:37,063 put in 207 00:17:37,067 --> 00:17:43,002 a 3rd and 4th request for census and they limit the amount of data that they will 208 00:17:43,006 --> 00:17:47,004 actually provide you so I had to be really selective in what and I did and did not 209 00:17:47,015 --> 00:17:52,040 request so I kind of had to focus. On. 210 00:18:00,014 --> 00:18:01,083 Oh yeah there's always 211 00:18:01,084 --> 00:18:05,055 a possibility that there's something else going on but in terms of what I could 212 00:18:06,012 --> 00:18:11,013 realistically get census to provide for me this is they wouldn't even give me half 213 00:18:11,014 --> 00:18:16,053 of what I asked for. Anyone else.