Managing Data Samples And Stats: Fill & Download for Free

IIM AhmedabadMumbai16th February Afternoon slotProfile:10/12/Graduation : 94.14/94.40/81.03Graduation: BSc Statistics,St. Xavier’s, MumbaiWork Experience: Aon Consulting (10 Months at the time of Interview)This was my 4th interview, and i had 3 interviews in the same week! (IIFT, IIM - Lucknow, IIM - Ahmedabad)AWT:Time: 20 minutesIt was a passage about how the law against ragging in Indian colleges should be revoked, as ragging was a way seniors and juniors mingled. There were a series of questions that followed the passage.Personal Interview:I was 5th in my panel, and entered at about 2:45 pm.There were two panelists, both middle aged men, professors. P1 looked a little disinterested :PI was called in, and greeted them both.P2: So Meghna, can you tell us a little about yourself?Spoke about where and what i studied, how i work in Human Resources Consulting (compensation practice) and how long it has been there.P2: Okay, what is compensation consulting? could you explain it to us?Me: Sir, we advise clients how best to pay their employees, and this we do by compensation benchmarking, etc, etc (explained the whole process hereafter)A few follow up questions from P1 about the process.P2: How do you think Statistics helped you have an edge with respect to the job?(i was pretty stumped, did not really use any tools i’d learnt in college)Me: Sir i think, i understand measures of central tendencies well, hence it becomes easier for me to produce compensation benchmarking reports.P1: See, averages anyone can compute, what else?Me: (trying to think of something that made sense) Sir, i ensure that the sample is representative in nature. And all companies data is well represented.P1: Okay, How do you ensure that?Me: Sir internally, we have a limit of a certain percentage of each sample to be of one company, and hence i ensure the same.P1: Can you write a formula where that might cause bias?Me: (absolutely confused) Sir, I'm sorry, i do not understand what formula you are referring to.P1: Do you think this method is foolproof? Dont you think stratified random sampling is better?Me: Yes sir, 95% cases it is, but sometimes it does cause an issue, and hence some clients request unsampled data. Wrt to Stratified sampling, it is very time consuming, and not very cost effective sir.P2: Why do they ask for unsampled ?Me: Sir these firms look more at headcount comparison than actual compensation comparison, as they perform mass recruitment, especially at lower levels.P2: was not completely convinced, but said okay.P1: First you said this is the method you use, now you're saying your method is wrong, and clients are dissatisfied.Me: No Sir, i said some large clients prefer the unsampled route, however largely our clients are okay with sampled data, as they want equal representation.P2: (Smiling) Yes, Yes, its fine.P2: So Meghna, what are some statistical tools used by humans?( i thought he meant database management tools)Me : Sir, R programming could be one.P2: No, no, like you studied concepts right, so any of those?Me: (so many, could not think of one) Sir, Big Data is a phenomenon floating around.P2: Any other core tool?Me: Sir, ANOVA is used extensively, like in biological sciences for Clinical Trials.P1: Okay so we don’t know anything about ANOVA or Clinical trials, can you explain it to us?Me: Sure Sir. i spoke about the placebo sugar pill and the test drug being administered to two groups and the mean effect of the two drugs is compared. (I screwed up here saying control group gets the test drug which was completely stupid! realised after the interview x_x )P1: can you write down the testing part for it?Me: okay sir, so we would have Placebo and Test drug, and administer different levels of it, say 50 ml and 100ml.P1: no, i want only 50 ml.Me: (i realised with only one level, we could use the T-test) Sir, in that case you can use a simple T-Test of means, since there is only one level.P1: Oh wow, first you started with Big data, then anova, now you are saying t-test!Me: No sir, i assumed we would administer multiple levels, hence.P1: Anyway, yes, continue the testing.Me: Wrote the null and alternate hypothesis. P1 asked me to draw the T distribution curve. Drew it. Asked me about Type 1 error. stopped me halfway when i was writing the procedure-P1: Okay, what distribution is the null hypothesis?Me: (confused) sir, that depends on the sample size, if it is large enough, it will tend to normal.P1: Okay. Now you said Big Data. What is this big data? what is big? how do they even define what is big?Me: (Stumped, i had no clue why it is called big, although had read a little) Sir, i read somewhere that it is data that cannot fit onto an excel sheet. (cant believe i said that)P1: Haha, that depends on the version of excel. No, why Else?Me: Sir maybe 1,00,000 and above data points in some cases?P1: So <100000 is medium data?? (how stupid of me)Me: Sir i guess it is largely contextual. Hence in some cases it might be safe to say 1,00,000.P1: So how is it even used, this big data?Me: Sir, in our country especially, we find it extremely difficult to get representative data, and sometimes a sample might not be completely adequate. However big data gives us a broader prospective, is more representative in nature, and its facets are a lot closer to the population parameters.P1: (not convinced) So your saying CENSUS is the best?Me: No Sir, larger data which is now readily available due to technology, and hence the population parameters are better furnished.P1: What Population Parameters?Me: could be anything you define - mean, etc.P1: Okay. So why do you want to do an MBA? you have only 10 months work ex.Me: Sir, the curriculum is perfect as it teaches you finance, marketing and HR, something that is not available in any other form, also the leadership skill development i think is brilliant. I also want to advance my corporate career.P1: (I think he was very sarcastic) But you seem to have learnt all the stats and know all the HR already, why do an MBA then?Me: No Sir, I have a lot to learn. i have spent only 10 months and do not know everything i need to about HR.P1: Okay Meghna, we are done. Thank you for being here.(At that moment i looked down at the toffee bowl. Haha)P1 continues - and you can take a toffee. all the best!Me: Thank you Sir.Verdict: ConvertedI guess P1 did try to grill me. I genuinely felt like that did not go well at all. In fact, i was guessing i would be rejected due to some mistakes i made.However,i converted and i am now joining the best b-school in the country! :)

Clearly it depends on exactly what you wish to research. The target is related to humans, and is very likely to have a subjective element because strictly objective data is often better obtained mechanically or through other sources with less possibility of human bias or cost.A Questionnaire can imply a qualitative interview with a set of themes, but I shall deal with the question on the understanding that we are looking at a quantitative situation as that is a far more common situation.The UK Census is one example of the highest quality questionnaire based data collection, is largely objective in terms of the data needed (eg how many people in the dwelling on a particular night), but there are still errors mainly due to non response and mistakes on the part of administration, analysis, respondents and interviewer. Statistical error is minimal because the sample size in 2011 was 94 per cent of the population and was estimated with 95 per cent confidence to be accurate to within +/- 85,000 (0.15 per cent). So that is what is possible with an enormous sample and a National scale budget.Because of its significance as a baseline from which to estimate things like demand for public services and infrastructure, Governments are usually willing to spend a lot on this particular form of questionnaire data collection. The results are cross matched with other sources of data (four example tax records or data from utilities) as a check. The ONS also does regular surveys using the much more common sampling techniques.In normal practice within usual research budget restraints, a sample is usually taken from a larger population, for example the opinions of 1,000 respondents randomly chosen can be fairly confidently taken to represent the opinions of the whole population, even one counted in tens of millions! The stats behind that statement that are well worth following up on.Sampling is one aspect of questionnaire administration which is crucial, and the other key factor in analysis is to be aware of the statistical validity of the data. For a particularly specialist subject it may be possible to contact all the people in the pool of potential respondents and conduct in depth surveys with each one. For issues that seek to understand what tens of millions of people think, then sampling of a fraction of the population is the only possibility, and for practical reasons the answers to a limited number of questions may be reduced to a yes/no or a scale. Obviously the quality (in the sense of type and detail) of information will vary according to the method chosen.One point to note is that all questionnaire based research does is to measure the response to the particular question being asked. Change one word or the even tone of a question and the response may vary significantly. Interviewers might be giving unconscious cues, so that is a good reason to use trained staff, and also to consider the character of the target audience. It may be that the demographics of the interviewers introduce a bias in data collection.So to answer the question more directly, merits of a questionnaires based research method are:1. Subjective data can be obtained which would otherwise be impossible to gather.2. Objective data which is otherwise unobtainable can be gathered subject to human errors (see below).3. Methodology is relatively well developed and understood.4. Expert samp!ing and administration and analysius can deliver statistically reliable results for large populations at (relatively) low cost .5. Very often, the same or similar questions have been researched in other times and locations, providing “free” points of reference for changes and comparisons between cultures and periods of time. A desk study may turn up very usefu! data which has already been collected on the same subject.6. Questionnaires can be international in scope, bypassing problems with other forms of data gathering like differences in reporting and recording.7. Many QC accredited firms already exist with the capacity to do fieldwork, making outsourcing relatively easy. Some of these companies can over pre-vetted and demographically sorted “panels” which means a quick turnaround.Demerits are:The technical difficulty of designing a practical set of questions which are not too subject to wordiness, ambiguity, respondent or interviewer boredom, general comprehension issues and deliverable within an acceptab!e time frame (as a very broad rule if thumb 20 minutes is generally considered to be a maximum before quality deteriorates). Questionnaire design and tbe analysis of results is a job for experts.Cost. Research budgets may not be generous enough to support good quality sampling and the the neccessary numbers.Errors and bias are difficult to eliminate.“Panels” (as in point 7 above) are a double edged tool, being non-random in the strict sense, but fast and controllable and can be adjusted for calibration purposes.Subjective subjects subject to change from day to day may deliver results which are out of date before even being published. (Time series data is often very useful for tracking trends or might even be the sole objective OTOH).Data only represents responses to the actual questions asked, it is very easy to read too much into it.International, interregional and cultural and geographical differences right down to local level can affect results if sampling is not carefully managed. (Although those differences might in themselves be of great significance to the aims if the research!)Apologies for not including useful links but my tablet app often loses my work when I switch windows to find the references. Google is your friend.

The meme is that 9 out of 10 startups are not successful. This means that "one in ten" is highly successful.3 startups fail, have gone bankrupt or business were closed3 startups are worse than average, aren’t very profitable, returning less than the invested capital, but are still active3 startups are a moderate success, are profitable, returning just the invested capital, were not acquired yet and are still active1 highly successful startup that will pay the investor a multiple return on all of his 10 investments; that startup had an IPO or were a target of a significant acquisitionEstimates of firm success rates range (...) to the widely acknowledged one in ten of the National Venture Capitalists Association.http://www.effectuation.org/research-papers/failing-firms-and-successful-entrepreneurs-serial-entrepreneurship-temporal-portfoliDATASmall business vs, VC seed startups vs. incubator/business accelerator performance stats data:Do Startup Seed Funding Programs Really Work?According to Imran Ghory, out of the 144 companies started via Y-Combinator’s nine programs - Y-Combinator’s results (March 2010):22 percent were failures57 percent are still independently operating29 percent got further funding post-Y-Combinator10 percent were acquiredDavid Cohen’s TechStars, has revealed its track record. TechStars’ results of their four programs in the last three years (March 2010):10 percent failed69 percent are still operating as independent companies56 percent got funding post-TechStars14 percent were acquiredhttp://www.kikabink.com/news/do-startup-seed-funding-programs-really-work/Investor performance in Israeli high-techIVC Online database, 2,462 Israeli hightech firms – the entire high-tech start-up population of the years 1991-2000 – were examined. The "performance" of the entire Israeli start-up sample based on the current (September, 2006) status of the portfolio companies:46 percent failure rate (2006)15-16 success rate (2006)12 percent M&As (2006)4 percent IPOs (2006)foreign CVCs success rate 34 percent, local CVCs 26 percentforeign VCs success rate 32 percent, local VCs 26 percentbusiness angels 24 percenttechnological incubators (531 start-ups) 3.5 percent success rateThe technological incubators have an extremely low success rate and an enormous failure rate. This could mostly reflect three factors – (1) that incubators invest at the pre-seed and seed stages and therefore the companies are inherently more risky; (2) that only low quality and/or very risky startups approach incubators and (3) an inefficient operation and business model. Incubator graduates that attract investment from VCs or CVCs have improved performance (10.8 percent and 10.5 percent, respectively). These results, however, are still well below average.This appeared in the Israel Venture Capital & Private Equity Journal (IVCJ). IVC Research Center publishes the Israel Venture Capital & Private Equity Journal, a quarterly review of trends and developments in the Israeli-related venture capital industry.http://www.altassets.com/private-equity-knowledge-bank/country-focus/article/nz11093.htmlSuccess rateA study of companies funded by VCs during the 1990s indicated that 14% of the companies went public and another 11% were acquired.National Venture Capitalists Association http://www.nvca.org/index.php?option=com_docman&task=doc_download&gid=673&Itemid=93Failure Increases Chances Of Success:Skill vs. Luck in Entrepreneurship and Venture Capital: Evidence from Serial Entrepreneurs (2006)18% chance of succeeding: First-time entrepreneurs20% chance of succeeding: Entrepreneurs who have previously failed30% chance of succeeding: Entrepreneurs who succeeded in a prior ventureHarvard Business School; Harvard University - Entrepreneurial Management Unit; National Bureau of Economic Research (NBER); European Corporate Governance Institute (ECGI)http://papers.ssrn.com/sol3/papers.cfm?abstract_id=933932http://onstartups.com/tabid/3339/bid/79/Six-Interesting-Stats-About-Startup-Success.aspxThe Development of Venture-Capital-Backed and Independent Companies21% aggregate failure rate for all (4841) companies (2004)53% failure rate for all (895) companies with external equity financing (2004)Of 895 venture capital-financed companies, only 420 were still active in 2004 - 50% of all financed companies failed, independent of their sub-sector origin.p. 154, chapter 10An Empirical Study Among Germany's Internet and E-Commerce Start-upshttp://www.emeraldinsight.com/books.htm?chapterid=1795470&show=abstracthttp://books.google.de/books?id=kmWCPJGGHAYCOn the Lifecycle Dynamics of Venture-Capital- and Non-Venture-Capital-Financed FirmsThey used a new data set that tracks U.S. firms from their birth over two decades, as collected by the U.S. Census Bureau, to examine the life cycle dynamics of VC- and non-VC financed firms, and to shed light on VC incentives. (...) It is possible that VCs show lower failure rates simply because some failures are disguised as acquisitions. We examine these possibilities in the data by looking at the failure dynamics and comparing VC to non-VC financed firms at the time of failure. -- Longitudinal Business Database between 1981 and 2001-- The matched sample contains 7,632 VC-financed firms and 7,632 non-VC-financed firms.11% failure rate (firms born 1981-1987)(VC-financed)(age 5 years)15% failure rate (firms born 1988-1994)(VC-financed)(age 5 years)11% failure rate (firms born 1995-2001)(VC-financed)(age 5 years)20%-25% failure rate (1981-1987)(VC-financed)(age 8-10 years)26%-31% failure rate (1988-1994)(VC-financed)(age 8-10 years)40% failure rate (non-VC-financed)(1981-1987)(age 5 years)44% failure rate (non-VC-financed)(1988-1994)(age 5 years)37% failure rate (non-VC-financed)(1995-2001)(age 5 years)p. 32http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1231698VC Failure Rate20 percent failure rate (32 investments) (2007)40 percent success rate with 5X+ returns (11 deals) (2007)Fred Wilson (Venture Investor) of Union Square Ventures failure rate as a venture capitalist of 17 years.http://techcrunch.com/2007/11/30/vcs-whats-your-failure-rate/Incubating30 percent failure rate (2001)70 percent success rate?Manoa Innovation Center, Hawaii; founded 1993 - http://www.htdc.org/http://www.allbusiness.com/business-planning/815336-1.htmlIncubating65-70 percent success rate (2010)30-35 percent failure rate?300 startups at Sunnyvale Plug and Play Tech Center (Silicon Valley) - Compared to an 80 percent failure rate among unsupported startups, according to the company's COO Kayvan Baroumand. (2010) - http://www.plugandplaytechcenter.com/http://www.gtlaw.com/portalresource/lookup/wosid/contentpilot-core-1602-11806/pdfCopy.name=/Daily%20Journal%20-%20Todd%20Rumberger%20-%20Lawyers%20Court%20Valley%20Startups%20(Unlimited%20Linking).pdfIncubating80 percent remain in business or have successfully merged (2005)20 percent failure rate?The Tacoma’s William M. Factory Small Business Incubator (WFSBI) has graduated more than 200 companies. "We have a very strict selection process," Colleen Barta, business development manager at the William M. Factory incubator says. "Otherwise, our 78 percent success rate could easily be a 78 percent failure rate." (2005)http://www.nbia.org/success_stories/awards/2005/william.phpIncubating"We also know that businesses that start in incubators have an 80 percent success rate" while those without such support have an 80 percent failure rate, Mike Pfeiffer, executive director of the South Park Economic and Development Corp. said. (1997)http://www.allbusiness.com/north-america/united-states-california-metro-areas/649632-1.htmlIncubatingThe success rate for businesses starting at the New River Valley Competitiveness Center has been estimated at 80 percent, compared to an 80 percent failure rate for those that start without the benefit of a business incubator. (2010)http://www.southwesttimes.com/news.php?id=4857Incubating"We have a 70-to-80% success rate rather than a 70-to-80% failure rate,” according to Susan Matlock, president of the Entrepreneurial Center, a small business incubator located in downtown Birmingham, Alabama. Matlock describes the businesses in the center as a mixture of service, light manufacturing and software.http://www.frbatlanta.org/pubs/partners/partners-vol_10_no_3-alabamas_small_business_incubators.cfmIncubating75-80 percent success rate of startup self-sustainability where they can exit the Incubator (2006)20-25 percent failure rate?Since inception in 2000, the maxum Business Incubator has supported the development of 30 companies. According to Jill Sawers, Incubator Manager. Internationally statistics show that about 60 percent of business start-ups fail within the first five years. Locally, anecdotal evidence indicates that the numbers are much higher - closer to an 80 percent failure rate. (2006)http://www.theinnovationhub.com/newsbits/vol5no4/news02.cfmIncubating85 percent success rate (2010)15 percent failure rate?"(...) failures occur if the technology or science doesn't prove out," John Korpela, director of the Business Innovation Network said. The business incubators run by Montgomery County's Department of Economic Development is home to as many as 30 start-ups and fledgling businesses.http://www.washingtonpost.com/wp dyn/content/article/2010/05/21/AR2010052104051.htmlFailure And Success Rates Of ~200 Web 2.0 Internet Startup CompaniesThe editgrid spreadsheet below contains 194 of the ~200 websites referred to within Ludwig Gatzke's popular Web 2.0 logo collage. This was posted in summer 2009.http://tomuse.com/internet-startups-web-20-success-failure-rates/#commentshttp://www.flickr.com/photos/stabilo-boss/93136022/http://www.editgrid.com/publish/html/user/vizionquest/43104641/?sort=1&filter=1&plain_table_mode=1&bgcolor=%23ffffff&fgcolor=%23000000&version=2&frame_style=height:500px;width:500pxVC Backed Startup Success RateAccording to the data of an NVCA study that speaks to the success rate for VC backed companies, about 45% of VC backed companies exit. However, it’s important to note that some of the exits in the ‘acquired’ category aren’t very profitable, just returning invested capital (or even less). As a result, the real wins are likely to be a subset of the 45% number.My gut tells me that the right number is about 33%; 1 out of 3 VC backed companies has a profitable exit.http://www.markpeterdavis.com/getventure/2008/09/vc-backed-start.htmlSmall BusinessThe data come from a special tabulation by the Bureau of the Census produced for the Office of Advocacy of the U.S. Small Business Administration.http://smallbiztrends.com/2008/04/startup-failure-rates.htmlSmall Business• First Year – 85% success• Second Year – 70% success• Third year – 62% success• Fourth Year – 55% success• Fifth Year 50% successDavid Birch, a former head of a research company investigated small businesses and formulated the above percentage rates for small business success in the first through tenth yearhttp://www.usatoday.com/money/smallbusiness/columnist/abrams/2004-05-06-success_x.htmSmall BusinessSeven out of 10 new employer firms survive at least two years, and about half survive five years. Findings do not differ greatly across industry sectors.U.S. Small Business Administration Office of Advocacy, September 2009; Survival and Longevity in the Business Employment Dynamics Database, Monthly Labor Review, May 2005. Redefining Business Success: Distinguishing Between Closure and Failure, Small Business Economics, August 2003.http://www.score.org/small_biz_stats.htmlSmall BusinessThe 2006 statistics from the Small Business Administration (SBA) show that "two-thirds of new employer establishments survive at lease two years, and 44 percent survive at least four years." "as a general rule of thumb, new employer businesses have a 50/50 chance of surviving for five years or more."http://www.businessknowhow.com/startup/business-failure.htmSmall Business vs, IncubatingPeople quote statistics about business failure rates (often incorrectly) all the time. The U.S. Small Business Administration has found that after four years, only 44 percent of firms remain in business.* Contrast that with what NBIA historically has found about incubator graduates: among incubation programs whose average age was nearly ten years, 87 percent of graduate firms were still in business.**Clearly, there is a huge difference between the survival rate of incubator graduates and the survival rate of small businesses in general. “If we could demonstrate that twice as many firms are still in business four years after graduating from an incubation program, that’s effectively doubling the success rate and bringing credibility to the incubation industry,” Skip Farrar, manager of business development at Southern California Edison, a public utility in California, says. “But you can’t figure that out until you know how long a graduate remains in business – its persistency in the market.”* U.S. Small Business Administration (SBA), “Frequently Asked Questions,” www.sba.gov/advo. Accessed on March 9, 2007.** Molnar, Lawrence A., Donald R. Grimes, Jack Edelstein, Rocco De Pietro, Hugh Sherman, Dinah Adkins and Lou Tornatzky, Business Incubation Works. Athens, Ohio: NBIA Publications, 1997. http://www.nbia.org/impact/what.php