HIGHLIGHTS

INTRODUCTION:

Cyber Citizens for Justice, Inc. [CCFJ] surveyed owners of residences in Florida Homeowners' Associations [HOAs] and affiliates of such HOAs in July and August of 2008.  The survey solicited respondents’ views about 14 possible legislative reforms.  1,033 usable responses were collected via the Internet. Six volunteers entered the usable responses into a standard database and forwarded it as an e-mail attachment to a seventh volunteer with survey analysis expertise.  Those volunteers are:  Kathleen d’Angelo-Frazier, Kandy Fancher, Joan Panrell, Joyce Sasser, Alan Sintef, Angela Stewart, Sue Sosa-Vega and David Goldenberg.   Dr. Goldenberg merged the databases the other volunteers submitted in a standard format into a master file and then analyzed its contents to produce this report.

You can access the tables in this report by clicking on the indicated link.  Links to the summary tables appear at the end of the discussion of the table in question.  The links to the detailed tables appear in the left column of a list following this highlights section.

The 19 tables in this report were developed as Microsoft Excel spreadsheets.  For technical reasons those tables lack the scrolling capability available in the original Excel versions.  If that deficiency makes them too difficult to view comfortably, please click here to access the tables in their original Excel format with the horizontal and vertical scrolling features active.  

Numbers in brackets [#] in the tables mean one of two things.  If they follow the statement of a question or issue in a summary table, the number represents the order the issue was raised in the questionnaire and refers you to the corresponding detailed table with the same number for the complete analysis of that issue.  Otherwise numbers in brackets refer you to a footnote in the list of footnotes appearing after the last detailed table.

Dr. Goldenberg’s analysis initially generated 14 tables with a standard heading and stub and a narrative set of conclusions at the top.  Each of those 14 tables detailed the findings about a specific possible legislative reform from six distinct viewpoints.  Those six viewpoints were an overall result as well as according to five different traits of respondents. 

Five traits of respondents were used to analyze the data.  They were:

• ‘Interest” or role of each respondent — that is, an owner, a past or present member of the board of directors of a Florida HOA, an attorney, a community association manager [CAM], etc. 

• “Gender”  

• “CCFJ membership or non-membership   

• “Region" of Florida

• “Remarks” — whether or not a respondent stated a priority for a particular piece of legislation and/or did or did not provide a comment was the last trait considered.

The following four highlights presented come from summaries of data from the 14 detailed tables. 

OVERALL % YES VS. NO TO ISSUES 1 – 14.

Summary Table 1 lists the overall results of respondents’ views of 14 different legislative reforms in descending order of favorable votes received.   The key finding here is that every potential legislative reform examined in the survey was deemed desirable.  Affirmative votes ranged from “only” 75.1% to 96.9%.   Those are notably high ratings.  The sizes of those affirmative votes are one indication of respondents’ priorities on these issues.

Seven of the 14 potential reforms had ratings in excess of 93.0%.   Three other potential reforms garnered favorable ratings of 90.9% or more but less than 92.1%.   These ratings are both notably high and remarkably consistent.

Four potential reforms had somewhat less positive ratings.  They concerned: Election Reforms, Education for Board Members, Shutters, and Rental Restrictions.   These four items are the “hottest” issues among respondents.    

SUMMARY TABLE 1: OVERALL % YES VS. NO TO ISSUES 1-14

CLUSTERING OF VOTES

Summary Table 2 deals with “clustering of votes” or block voting.  Almost 55% of respondents voted either YES on all 14 possible legislative reforms or NO on all 14.  53.7% voted YES across the board while only 1.1% voted NO to everything. 

Some 42.8% of all respondents voted YES most of the time, eight to 12 times out of a possible 14.  But only 1.5% of the respondents usually voted NO.  

Block voting varied considerably across respondents’ traits as one would expect.  Respondents who were CCFJ members and residents of either the Northern or West Coast region were far more likely to vote in favor of all 14 proposed reforms.  Conversely, respondents who were board members, lawyers, or community association managers were significantly more likely to block vote against all 14 proposals.

Clustering or block voting may indicate the presence of a bias or predisposition to see things as either black or white with no intervening shades of grey.  

Almost 44% of all respondents split their votes in one of three fashions.  Just under 1% voted for seven proposals and against the other seven. Some 42.8% voted in favor of eight to 13 proposals and against one to six others.  Conversely, 1.5% of respondents voted against eight to 13 proposed legislative reforms and for one to six others.

SUMMARY TABLE 2: CLUSTERING OF VOTES

BIASES?

Summary Table 3: Presence or Absence of Bias within Respondent Traits signals the direction, intensity and consistency of biases discerned. For those unfamiliar with statistics, the following paragraph explains how this table was created and why its findings matter.

Each of the detailed tables covering a given potential legislative reform used boldfacing to mark those percent YES or NO answers in each respondent trait which differed “significantly” from the overall average value.  If the percent YES or NO in a respondent trait section was three or more standard deviations away from the overall average, then it was deemed “significant.”  The odds of a particular value actually being the same as the overall average when that value is three standard deviations away from that average is 1 to 332.  A significant deviation from the overall average can, of course, be greater than or less than an overall average.  Three cases are possible with an overall averages for YES and another for NO.

Case 1: respondents in a given trait category have a significantly higher YES value and  also a significantly lower NO value than the corresponding overall averages.   Such instances were noted in the appropriate cells of Summary Table 3 with a “B,” for beyond the overall averages.

Case 2:  respondents in a given trait category have a significantly lower YES value and also a significantly higher NO value than the corresponding overall averages.  Such instances were noted in the appropriate cells of Summary Table 3 with a “W,” for within the overall averages.

Case 3:  respondents in a given trait category offered YES and NO answers which did not significantly differ from the overall averages.  Such instances were indicated by leaving the appropriate cells of Summary Table 3 blank.

Eight of the 20 subgroups of respondent traits exhibited eight or more significant deviations from the overall averages.  Five of those subgroups had consistent W biases.  They were CAMs with 9, Comment Only with 11, Lawyers and Other tied with 12 each, and Board Members with 13.   The other three subgroups had consistent B biases with residents of the Northern Region having 10 and CCFJ members and residents of the West Coast tied at eight apiece. Given the odds of each deviation being significant coupled with the internal consistency of the pattern of apparent bias in any of these subgroups, the laws of probability warn that the odds of any of these groups actually being unbiased is so vanishingly small as to be quite literally infinitesimal, specifically [(1/333)^8] for CCFJ members and residents of the West Coast.

Only two of the 20 subgroups of respondent traits had YES and NO percentages to all 14 questions which essentially were the same as the overall average and so apparently were unbiased.  Those subgroups consisted of non-members of CCFJ and residents in the Central Region of Florida.

Another subgroup, Females, had just one B and one W which offset one another in signaling the presence of a bias.

Three of the 20 subgroups of respondent traits had only one significantly different answer.  Males had a B but residents of the Southern Region and respondents who did not indicate a priority and did not provide a comment had one W each.  

The Owners subgroup had three biased answers and, as might be expected, those consistently were B.

The above information warns that responses of Board Members, Lawyers, and C.A.M.s to these 14 proposed legislative reforms assuredly do not accurately reflect the views of other homeowners in Florida HOAs.  This is a crucial finding in the light of previously circulated research [2002 National Coalition for Homeowner Rights, 2005 Community Associations Institute/Zogby International, CALL’s 2005 Florida Community Living Survey  which relied heavily on the advice of HOA Board Members, and 2006 Survey by American Homeowners Resource]. 

This survey could not examine the causes of the biases detected.  Given the absence of data, it’s advisable not to speculate about the cause(s) of the biases found.  One should, however, bear in mind that differences in experience, professional training, and some personal characteristics [such as altruism or selfishness, apathy or energy level] may, and when data is available usually do, better explain any biases found than a simplistic assumption of malice or stupidity.

Two additional comments about bias.   First, everyone is somewhat biased.  What matters is the relevance of the bias and its consistency and intensity.   This survey found some relevant, consistent and intense biases.  Second, having a bias is not the same as being wrong.  It only means one is somewhat different than the average on an issue or a series of issues.

SUMMARY TABLE 3: PRESENCE or ABSENCE OF BIAS WITHIN RESPONDENT TRAITS

REGIONALITY  

Some information generated by this survey indicates that problems with boards of directors of Florida HOAs are common — that is, an epidemic rather than isolated incidents or endemic as some claim.  Responses to this survey came from all over the state in rough proportion to both population and the local popularity of HOAs.  That’s one indication of an epidemic.  Another is the relatively high response rates from all regions. A third signal derives from the fact that the five regions’ YES votes were statistically significantly higher than the overall averages 27.1% of the time [19 incidents out of 70 possibilities {5 regions * 14 issues}] but only once or 5.0% of the time was there a response statistically significantly lower than the overall value. 

The five regions are not alike according to the levels and frequencies of statistically significant responses differing from the overall average.  The North is considerably more positive on almost every issue than any other region, especially the South.  The responses from the Central region are consistently closer to the overall averages than the other regions.

SUMMARY TABLE 4: REGIONALITY

DETAILED TABLES

The following section presents 14 tables in a standard format plus a list of their accompanying footnotes. Each of the tables details the responses to one particular question. The tables are sequenced in the same order as the question was asked in the survey questionnaire. Just below is a two-column list. The 14 questions asked appear in the right-hand column with a link to the corresponding table just to the left of each question. Click the underlined “Table #” in the box to the left of the question and the appropriate table will appear. 

 

TABLE   1	WILLING TO PAY $4 PER HOUSEHOLD PER YEAR TO FUND AN INDEPENDENT REGULATORY AGENCY TO PROTECT HOA OWNERS (#1)
TABLE   2	CREATE A HOA OMBUDSMAN'S OFFICE LIKE THE ONE FOR CONDOMINIUMS (#2)
TABLE   3	ENACT HOA ELECTION REFORMS IN LINE WITH CONDOMINIUMS (#3)
TABLE   4	HOLD DIRECTORS PERSONALLY ACCOUNTABLE FOR THEIR ACTS AND BAR FELONS FROM SERVING AS DIRECTORS PLUS OTHER RESTRICTIONS (#4)
TABLE   5	REQUIRE DIRECTORS TO BE DEEDED HOMEOWNERS & LIMIT THEM TO A 2-YEAR STAGGERED TERM ENDING AT ANNUAL MEETINGS PLUS OTHER RESTRICTIONS (#5)
TABLE   6	BAR ALL KINDS OF FINANCIAL COMPENSATION FOR DIRECTORS EXCEPT FOR REPAYMENT OF PRE-AUTHORIZED AND DOCUMENTED OUT-OF-POCKET EXPENSES (#6)
TABLE   7	EDUCATION / TRAINING REQUIREMENT FOR HOA DIRECTORS AND  HOMEOWNERS WITH CERTIFICATION AS PRECONDITION FOR BOARD SERVICE (#7)
TABLE   8	CLARIFY RULES FOR RECORD REQUESTS AND ENFORCE THOSE RULES (#8)
TABLE   9	IMPOSE NOTIFICATION SAFEGUARDS BEFORE CREATING OR FORECLOSING A LIEN ON A HOA RESIDENCE (#9)
TABLE 10	PROTECT HOA OWNERS' RIGHTS TO ATTEND AND SPEAK AT MEETINGS OF THE BOARD OF DIRECTORS (#10)
TABLE 11	BAR HOA BOARDS FROM INTRODUCING RENTAL RESTRICTIONS NOT ALREADY IN THE DOCUMENTS UNLESS A SUPERMAJORITY OF HOMEOWNERS APPROVES (#11)
TABLE 12	BAR HOA BOARDS FROM REQUIRING OR LIMITING INSTALLATION OF  HURRICANE SHUTTERS OR LIMITING OPENING OR CLOSING APPROVED SHUTTERS (#12)
TABLE 13	BAR DEVELOPERS FROM CHANGING HOA DEED RESTRICTIONS WITHOUT HOMEOWNERS' APPROVAL PLUS OTHER PROTECTIONS (#13)
TABLE 14	REVITALIZATION CONTROL AND ENFORCEMENT TO REQUIRE WRITTEN NOTICE AND WRITTEN APPROVAL BY AT LEAST 75% OF DEEDED OWNERS (#14)

TECHNICAL APPENDIX

Cyber Citizens for Justice, Inc. sponsored a survey of owners, residents and affiliates of homeowner associations (HOAs) in Florida.  (A copy of the questionnaire appears below.)  Affiliates included past and present members of boards of directors of Florida HOAs, attorneys, and community association managers (CAMs).   The survey ran from 7 July 2008 through 17 August 2008.  Responses were collected via the Internet.

The survey generated 1,033 usable responses out of 1,123 responses submitted.  Ninety responses were rejected for any of four distinct reasons.  48 responses were rejected because they were submitted anonymously in violation of an announced requirement that every respondent identify themselves.  Responses submitted after the deadline also were ignored.

Another 37 responses were rejected as duplicates.  Only one response per person or household was permissible.  When a duplicated submission was found, the more recent one replaced the earlier submission, provided, of course, that it met all the other requirements.

Three responses were rejected because they came from people who did not own a residence in a HOA in Florida although they lived in a HOA in another state.  The last two rejected responses were, depending on the reader’s viewpoint,  either gibberish or jokes.

This survey is a preliminary effort to learn the views of homeowners in and affiliates of Florida HOAs about potential legislative initiatives to improve conditions in Florida HOAs.  This survey represents the views of the respondents at the time it was conducted. While indicative of the views of the general population of Florida HOA owners and affiliates, this survey is not statistically representative of the entire population of people owning, living in or affiliated with Florida homeowners associations because it could not be based on a random sample.  Therefore it is not projectible to all members of that broader group at the time it was administered. A reliable, current and complete list of such individuals is essential if one is to have an affordable survey which reliably characterizes all of the relevant population rather than just the specific respondents. Alas, no such list exists.  Nevertheless, this survey is an important initial effort and its high degree of internal consistency enhances its credibility. 

The data collected formed a database in a Microsoft Excel spreadsheet.  The data on all 14 potential legislative initiatives were analyzed in a consistent manner with three widely used statistical tools available in Excel.  Those initiatives were assessed from an overall perspective as well as by five different traits of the respondents.  Those analytic tools were percentages, t-tests and Chi-squared tests.  T-tests were used to determine if any given  percentage response within a trait of respondents was or was not statistically significantly different from the overall average.  Determinations of such differences relied on a much higher standard than normally used commercially.  The  conventional criterion for a significant departure from an average is + 2 or more standard deviations from the average.  A value of two [2] standard deviations from the average has only 1 chance out of 20 of actually being the same as the average.  This survey raised that criterion to  + 3 standard deviations or less than 1 chance out of 333 of actually being the same as the average.  The + 3 standard deviations used is almost 17 times more rigorous than the one commonly employed.  That more stringent test was selected to minimize the possibility of drawing an incorrect conclusion from the data produced by the non-random sample. 

Chi-squared tests were used to determine if a particular trait of respondents was related to or explained part or all of the answers to a given question in a cross-tabulated table or, alternatively, if the answers were independent of the trait.  The respondent traits were: Interest, Gender, CCFJ membership or non-membership, Region, and Additional Remarks Made.  A table of the results of the Chi-squared tests is appended to help future researchers design more robust samples in the future and to focus on the more likely causal variables in those studies. 

For all practical purposes, the Interest of respondents was by far the most important explanatory variable.  It mattered in ten of the 14 issues.  Membership or non-membership in the Cyber Citizens for Justice was much weaker explanatory variable as it only mattered in four of the 14 issues mostly in conjunction with Interest.  Gender, Additional Remarks Made, and Region essentially were unimportant as explanatory variables.   Three important conclusions follow logically from this latter point. 

First, the problems of conflicting Interests and viewpoints exist throughout to Florida rather than being geographically localized phenomena.  Region only mattered on one issue and then in conjunction with three other traits of respondents. 

Second, the problems are independent of Gender save in two instances, once in conjunction with Interest alone and the second time in conjunction with Interest, CCFJ Membership and Region. 

Third, whether or not a respondent indicated which issues took priority with her/him or provided a comment only influenced the respondent’s vote on two issues, Willingness to fund a state regulatory agency of HOAs and Barring developers from changing deed restrictions without homeowners’ approval.  However, the priorities and comments given represent potentially valuable additional sources of information which will be explored later. 

Respondents’ age, income, length of residence in a Florida HOA, nationality, and/or education/profession are potential demographic variables which may influence respondents’ votes.  However, neither those demographic  variables nor a number of psychographic ones were explored in this survey.  Nor were such additional potential explanatory variables as a HOA’s age, size in terms of number of residents, history of trouble [complaints to the DBPR, law suits by or against the board, recalls, etc.], identity of the developer, identity of the management company, and so forth taken into account in this preliminary survey.

SUMMARY TABLE 5: CAUSAL RELATIONSHIPS FOUND

QUESTIONNAIRE FOR 2008 HOA REFORM SURVEY

***************************************************************************

END OF WRITTEN REPORT

THE SUMMARY USING THE ORIGINAL POSTING OF SURVEY:

SURVEY QUESTIONS